- Email: [email protected]
The Incremental Role of Magnetic Resonance Imaging for Prostate Cancer Staging before Radical Prostatectomy
EURURO-6970; No. of Pages 4 EUROPEAN UROLOGY XXX (2016) XXX–XXX
available at www.sciencedirect.com journal homepage: www.europeanurology.com
Brief Correspondence
The Incremental Role of Magnetic Resonance Imaging for Prostate Cancer Staging before Radical Prostatectomy Alessandro Morlacco a, Vidit Sharma a, Boyd R. Viers a, Laureano J. Rangel b, Rachel E. Carlson b, Adam T. Froemming c, R. Jeffrey Karnes a,* a
Department of Urology, Mayo Clinic in Rochester, MN, USA; b Department of Health Science Research, Mayo Clinic in Rochester, MN, USA; c Department of
Radiology, Mayo Clinic in Rochester, MN, USA
Article info
Abstract
Article history: Accepted August 5, 2016
In the present report we aimed to analyze the incremental value of preoperative magnetic resonance imaging (MRI), in addition to clinical variables and clinicallyderived nomograms, in predicting outcomes radical prostatectomy (RP). All Mayo Clinic RP patients who underwent preoperative 1.5-Tesla MRI with endo-rectal coil from 2003 to 2013 were identified. Clinical and histopathological variables were used to calculate Partin estimates and Cancer of the Prostate Risk Assessment (CAPRA) score. MRI results in terms of extracapsular extension (ECE), seminal vesicle invasion (SVI), and lymph-node invasion (N+) were recorded. Using RP pathology as gold standard, we developed multivariate logistic regression models based on clinical variables, Partin Tables, and CAPRA score, and assessed their predictive accuracy before and after the addition of MRI results. Five hundred and one patients were included. MRI + clinical models outperformed clinical-based models alone for all outcomes. Comparing Partin and Partin + MRI predictive models, the areas under the curve were 0.61 versus 0.73 for ECE, 0.75 versus 0.82 for SVI, and 0.82 versus 0.85 for N+. Comparing CAPRA and CAPRA + MRI models, the areas under the curve were 0.69 versus 0.77 for ECE, 0.75 versus 0.83 for SVI, and 0.82 versus 0.85 for N+. Our data show that MRI can improve clinical-based models in prediction of nonorgan confined disease, particularly for ECE and SVI. Patient summary: Magnetic resonance imaging, together with clinical information, can be useful in preoperative assessment before radical prostatectomy. # 2016 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Associate Editor: Giacomo Novara Keywords: MRI Staging Risk assessment Prostate cancer Prostatectomy Predictive models CAPRA score Partin Table
* Corresponding author. Mayo Clinic, Gonda Building 7-130, 200 First Street South West Rochester, MN 55905, USA. Tel. +1-507-266-9968; Fax: +1-507-284-4951. E-mail address: [email protected] (R.J. Karnes).
Prediction of adverse pathological features at surgery, in terms of extracapsular extension (ECE), seminal vesicle invasion (SVI), and node positivity (N+), is of particular interest when defining the surgical approach and the role of magnetic resonance imaging (MRI) in this setting is still a matter of debate. The European Association of Urology [1] and National Comprehensive Cancer Network guidelines [2] do not provide definitive indications for MRI use, simply
hypothesizing a role for this technique, especially in high-risk disease. A recent meta-analysis [3] concluded that MRI has a good specificity for T staging, while sensitivity is highly variable. As such, there remains a high-level of clinical uncertainty regarding the potential role of MRI, particularly when compared with conventional, clinically-based risk classification models. In the present study, we compared MRI with existing models (the Cancer of the Prostate Risk Assessment [CAPRA] score and the
http://dx.doi.org/10.1016/j.eururo.2016.08.015 0302-2838/# 2016 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Please cite this article in press as: Morlacco A, et al. The Incremental Role of Magnetic Resonance Imaging for Prostate Cancer Staging before Radical Prostatectomy. Eur Urol (2016), http://dx.doi.org/10.1016/j.eururo.2016.08.015
EURURO-6970; No. of Pages 4 2
EUROPEAN UROLOGY XXX (2016) XXX–XXX
Partin tables) [4,5] and analyzed if the addition of MRI information could improve these clinical models in terms of pathological outcomes at the time of radical prostatectomy (RP). All patients who underwent preoperative 1.5-Tesla MRI and subsequent RP with pelvic lymph node dissection at Mayo Clinic Rochester, MN, USA, between 2003 and 2013 were included in this retrospective, Institutional Review Board approved study. Detailed methods are reported in Supplementary Data 1. In brief, we used age, prostate-specific antigen (PSA) at diagnosis, clinical stage (digital rectal exam), primary and secondary Gleason grade, Gleason score, percentage of positive/total cores to calculate Partin Table estimates and CAPRA scores. MRI results for ECE, SVI, and nodal disease (N+) were collected. Pathological features (pT, pN, Gleason score, surgical margin status, SVI, and ECE) were recorded. We developed logistic
multivariable regression models including: clinical variables (PSA, clinical stage, percentage of involved cores/total cores, primary Gleason 4–5), Partin Table estimates (for each specific outcome) and CAPRA scores; MRI results (in terms of negative/positive exam for each outcome) were then added in each model and the multivariate modeling was reassessed. The predictive ability of each model was further compared developing receiver-operating characteristic curves and analyzing the area under the curve (AUC) before and after the addition of MRI. Five hundred and one patients were included in the final analysis. Demographic, clinical, and pathological features are shown in Supplementary Table 1. ECE, SVI, and N+ were present in 42.3%, 30.7%, and 16.0% of patients, while MRI results were positive for ECE in 147 patients (29.3%), for SVI in 83 patients (16.6%), and for N+ disease in 24 (4.8%). As shown in Table 1, the multivariate clinical models for
Table 1 – Multivariate modeling analysis for predictors of histological outcome ECE prediction
Without MRI OR
Clinical variables PSA % core involv. Age at surgery Clinical stage Primary Gleason 4/5 MRI ECE result Partin ECE MRI ECE result CAPRA score MRI ECE result
1.025 1.014 1.024 1.348 2.344 — 1.042 — 1.328 —
95% CI
1.000 1.005 0.990 1.124 1.374 — 1.019 — 1.191 —
SVI prediction
1.007 1.026 0.997 1.273 5.707 — 1.109 — 1.507 —
95% CI
0.983 1.015 0.959 1.048 2.820 — 1.075 — 1.332 —
N+ prediction
OR
0.054 0.003 0.165 0.001 0.002 — 0.0003 — <0.0001 —
1.021 1.015 1.016 1.220 4.145 2.379 1.023 7.522 1.244 5.248
95% CI
0.991 1.004 0.976 0.975 2.194 1.262 0.995 4.108 1.095 2.994
1.033 1.037 1.037 1.547 11.552 — 1.145 — 1.706 —
1.023 1.024 1.003 1.304 7.147 — 1.115 — 1.660 —
95% CI
0.996 1.009 0.954 1.025 2.066 — 1.074 — 1.415 —
1.051 1.039 1.054 1.658 24.726 — 1.157 — 1.948 —
p value
1.052 1.026 1.058 1.525 7.833 4.488 1.052 13.773 1.413 9.198
0.173 0.007 0.427 0.081 <0.0001 0.007 0.102 <.0001 0.0008 <0.0001
With MRI p value
OR
0.5572 <0.0001 0.8838 0.0148 <0.0001 — <0.0001 — <0.0001 —
1.010 1.027 0.984 1.152 5.073 7.024 1.098 7.582 1.408 8.854
95% CI
0.983 1.015 0.940 0.922 2.233 3.003 1.058 3.449 1.222 4.286
Without MRI OR
Clinical variables PSA % involv core Age at surgery Clinical stage Primary Gleason 4/5 MRI nodes result Partin nodes MRI nodes CAPRA score MRI nodes
p value
Without MRI OR
Clinical variables PSA % core involv. Age at surgery Clinical stage Primary Gleason 4/5 MRI SVI result Partin SVI MRI SVI result CAPRA score MRI SVI result
1.051 1.023 1.060 1.616 4.000 — 1.065 — 1.481 —
With MRI
p value
1.038 1.040 1.029 1.441 11.524 16.431 1.139 16.669 1.622 18.288
0.4721 <0.0001 0.4768 0.2130 0.0001 <.0001 <0.0001 <0.0001 <0.0001 <0.0001
With MRI p value
OR
0.0964 0.0012 0.9170 0.0304 0.0019 — <0.0001 — <0.0001 —
1.032 1.020 0.986 1.344 9.669 4.831 1.102 14.249 1.682 15.060
95% CI
0.996 1.003 0.928 1.014 2.263 1.320 1.056 3.141 1.392 4.348
p value
1.068 1.037 1.048 1.782 41.309 17.674 1.150 64.638 2.032 52.165
0.0794 0.0235 0.6558 0.0395 0.0022 0.0173 <0.0001 0.0006 <0.0001 <0.001
CAPRA = Cancer of the Prostate Risk Assessment; CI = confidence interval; ECE = extracapsular extension; MRI = magnetic resonance imaging; N+ = lymph-node invasion; OR = odds ratio; PSA = prostate-specific antigen; SVI = seminal vesicle invasion.
Please cite this article in press as: Morlacco A, et al. The Incremental Role of Magnetic Resonance Imaging for Prostate Cancer Staging before Radical Prostatectomy. Eur Urol (2016), http://dx.doi.org/10.1016/j.eururo.2016.08.015
EURURO-6970; No. of Pages 4 EUROPEAN UROLOGY XXX (2016) XXX–XXX
3
Fig. 1 – receiver operating characteristic (ROC) curves for extracapsular extension (ECE) and seminal vesicle invasion (SVI) prediction. CAPRA = Cancer of the Prostate Risk Assessment; MRI = magnetic resonance imaging.
prediction of ECE, SVI, and nodal disease included age, PSA, clinical stage, primary Gleason score 4 or 5, and percentage of involved cores/total cores. All of these, except for age and PSA, were significant predictors of outcome, with primary Gleason 4 or 5 being the strongest factor associated with each outcome (odds ratio [OR] 2.3, p < 0.01 for ECE; OR 5.7, p < 0.0001 for SVI; OR 7.147, p < 0.01 for N+ disease). When MRI results were added to each model, they showed a significant association with the outcomes (all p < 0.001). Combined MRI and each model outperformed individual models alone at all levels in the receiver-operating characteristic analysis (Supplementary Table 2). Comparing Partin and Partin + MRI predictive models, the AUC was 0.61 versus 0.73 for ECE, 0.75 versus 0.82 for SVI (Fig. 1), and 0.82 versus 0.85 for N+ (Supplementary Fig. 1). Comparing CAPRA and CAPRA + MRI models, the AUC was 0.69 versus 0.77 for ECE, 0.75 versus 0.83 for SVI (Fig. 1), 0.82 versus 0.85 for N+ (Supplementary Fig. 1). Our results suggest that MRI information can provide an added value when compared with clinical-based, Partin Table, or CAPRA score models alone for the prediction of adverse histopathological outcomes at RP. Based on our data and on the nature of our cohort, patients with doubtful clinical findings and intermediate to high risk disease could
benefit the most from preoperative MRI. Few studies have explored the potential additional value of MRI, and those that have are relatively underpowered. Feng et al [6] compared the predictive accuracy of MRI and clinical models (Partin Tables and Memorial Sloan Kettering Cancer Center nomogram) for ECE finding a minor improvement in diagnostic accuracy after addition of MRI to the model (AUC for Partin Tables and Memorial Sloan Kettering Cancer Center of 0.85 and 0.86, respectively, increased to 0.93 and 0.94 after addition of MRI). The magnitude of the difference they detected is likely lower than ours because they had a lower risk cohort and thus had lower rates of ECE, which would falsely inflate the AUC for models that predicted against ECE and underestimate the utility of an MRI. For instance, 95% of their cohort was either cT1c or cT2a, while only 66% of our core met these criteria. Gupta et al [7] compared the accuracy of 3-Tesla MRI and Partin Tables in predicting ECE. This study did not analyze the additional value of MRI to the Partin Tables, but rather compared those two tools in alternative models, finding a Partin accuracy similar to ours (AUC: 0.62) and a MRI AUC of 0.82. Again, however, the study population was small (60 patients) and with a clear preponderance of clinically low and intermediate-risk patients. Tay et al [8] used 3-Tesla
Please cite this article in press as: Morlacco A, et al. The Incremental Role of Magnetic Resonance Imaging for Prostate Cancer Staging before Radical Prostatectomy. Eur Urol (2016), http://dx.doi.org/10.1016/j.eururo.2016.08.015
EURURO-6970; No. of Pages 4 4
EUROPEAN UROLOGY XXX (2016) XXX–XXX
MRI for ECE prediction, finding an AUC of 0.69 for clinical model alone, 0.72 for clinical model + nonspecializedreading MRI and 0.91 for the two previous elements + specialized (dedicated radiologist) MRI reading. In our study, all the radiologists involved in final MRI readings were consultants with a consolidated experience in genitourinary pathology, and the incremental benefit was already evident at first reading. Our study has several limitations, including the retrospective nature, the high percentage of high-risk patients in the cohort, the long study period, and the lack of use of the Prostate Imaging Reporting and Data System system [9]. A detailed discussion of these aspects can be found in Supplementary Data 2. In conclusion, MRI provides additional predictive value to clinical-based models alone, but it is possible that our findings underestimate the true diagnostic accuracy of MRI and its added value. As such, prospective investigations with Prostate Imaging Reporting and Data System and multiparametric MRI use are warranted to clarify the ability of MRI to improve clinical prediction in this setting. Author contributions: R. Jeffrey Karnes had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Morlacco, Karnes, Sharma, Froemming. Acquisition of data: Morlacco, Carlson, Rangel, Sharma, Viers. Analysis and interpretation of data: Morlacco, Rangel, Carlson, Froemming, Karnes. Drafting of the manuscript: Morlacco, Karnes, Viers, Sharma, Froemming.
Funding/Support and role of the sponsor: This publication was made possible by Clinical and Translational Science Award Grant Number UL1 TR000135 from the National Center for Advancing Translational Sciences, a component of the National Institutes of Health. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of National Institutes of Health.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. eururo.2016.08.015. References [1] EAU 2015 Guidelines on Prostate Cancer. https://uroweb.org/ guideline/prostate-cancer/. Accessed November 28, 2015. [2] NCCN. NCCN Guidelines on Prostate Cancer.1.2016. https://www. nccn.org/store/login/login.aspx?ReturnURL=http://www.nccn.org/ professionals/physician_gls/pdf/prostate.pdf. [3] de Rooij M, Hamoen EHJ, 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. [4] Cooperberg MR, Pasta DJ, Elkin EP, et al. The University of California, San Francisco Cancer of the Prostate Risk Assessment score: A straightforward and reliable preoperative predictor of disease recurrence after radical prostatectomy. J Urol 2005;173:1938–42. [5] Eifler JB, Feng Z, Lin BM, et al. An updated prostate cancer staging nomogram (Partin tables) based on cases from 2006 to 2011. BJU Int 2013;111:22–9. [6] Feng TS, Sharif-Afshar AR, Wu J, et al. Multiparametric MRI improves
Critical revision of the manuscript for important intellectual content:
accuracy of clinical nomograms for predicting extracapsular exten-
Morlacco, Sharma, Viers, Rangel, Carlson, Froemming, Karnes.
sion of prostate cancer. Urology 2015;86:332–7.
Statistical analysis: Rangel, Carlson, Morlacco.
[7] Gupta RT, Faridi KF, Singh AA, et al. Comparing 3-T multiparametric
Obtaining funding: None.
MRI and the Partin tables to predict organ-confined prostate cancer
Administrative, technical, or material support: None.
after radical prostatectomy. Urol Oncol 2014;32:1292–9.
Supervision: Karnes, Froemming. Other: None.
[8] Tay KJ, Gupta RT, Brown AF, Silverman RK, Polascik TJ. Defining the incremental utility of prostate multiparametric magnetic resonance imaging at standard and specialized read in predicting extracapsular
Financial disclosures: R. Jeffrey Karnes certifies that all conflicts of interest, including specific financial interests and relationships and
extension of prostate cancer. Eur Urol 2016;70:211–3. [9] Radtke JP, Hadaschik B, Wolf M, et al. The impact of magnetic
affiliations relevant to the subject matter or materials discussed in the
resonance imaging on prediction of extraprostatic extension and
manuscript (eg, employment/affiliation, grants or funding, consultan-
prostatectomy outcome in low-, intermediate- and high-risk pros-
cies, honoraria, stock ownership or options, expert testimony, royalties,
tate cancer patients. Try to find a standard. J Endourol 2015;29:
or patents filed, received, or pending), are the following: None.
1396–405.
Please cite this article in press as: Morlacco A, et al. The Incremental Role of Magnetic Resonance Imaging for Prostate Cancer Staging before Radical Prostatectomy. Eur Urol (2016), http://dx.doi.org/10.1016/j.eururo.2016.08.015
available at www.sciencedirect.com journal homepage: www.europeanurology.com
Brief Correspondence
The Incremental Role of Magnetic Resonance Imaging for Prostate Cancer Staging before Radical Prostatectomy Alessandro Morlacco a, Vidit Sharma a, Boyd R. Viers a, Laureano J. Rangel b, Rachel E. Carlson b, Adam T. Froemming c, R. Jeffrey Karnes a,* a
Department of Urology, Mayo Clinic in Rochester, MN, USA; b Department of Health Science Research, Mayo Clinic in Rochester, MN, USA; c Department of
Radiology, Mayo Clinic in Rochester, MN, USA
Article info
Abstract
Article history: Accepted August 5, 2016
In the present report we aimed to analyze the incremental value of preoperative magnetic resonance imaging (MRI), in addition to clinical variables and clinicallyderived nomograms, in predicting outcomes radical prostatectomy (RP). All Mayo Clinic RP patients who underwent preoperative 1.5-Tesla MRI with endo-rectal coil from 2003 to 2013 were identified. Clinical and histopathological variables were used to calculate Partin estimates and Cancer of the Prostate Risk Assessment (CAPRA) score. MRI results in terms of extracapsular extension (ECE), seminal vesicle invasion (SVI), and lymph-node invasion (N+) were recorded. Using RP pathology as gold standard, we developed multivariate logistic regression models based on clinical variables, Partin Tables, and CAPRA score, and assessed their predictive accuracy before and after the addition of MRI results. Five hundred and one patients were included. MRI + clinical models outperformed clinical-based models alone for all outcomes. Comparing Partin and Partin + MRI predictive models, the areas under the curve were 0.61 versus 0.73 for ECE, 0.75 versus 0.82 for SVI, and 0.82 versus 0.85 for N+. Comparing CAPRA and CAPRA + MRI models, the areas under the curve were 0.69 versus 0.77 for ECE, 0.75 versus 0.83 for SVI, and 0.82 versus 0.85 for N+. Our data show that MRI can improve clinical-based models in prediction of nonorgan confined disease, particularly for ECE and SVI. Patient summary: Magnetic resonance imaging, together with clinical information, can be useful in preoperative assessment before radical prostatectomy. # 2016 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Associate Editor: Giacomo Novara Keywords: MRI Staging Risk assessment Prostate cancer Prostatectomy Predictive models CAPRA score Partin Table
* Corresponding author. Mayo Clinic, Gonda Building 7-130, 200 First Street South West Rochester, MN 55905, USA. Tel. +1-507-266-9968; Fax: +1-507-284-4951. E-mail address: [email protected] (R.J. Karnes).
Prediction of adverse pathological features at surgery, in terms of extracapsular extension (ECE), seminal vesicle invasion (SVI), and node positivity (N+), is of particular interest when defining the surgical approach and the role of magnetic resonance imaging (MRI) in this setting is still a matter of debate. The European Association of Urology [1] and National Comprehensive Cancer Network guidelines [2] do not provide definitive indications for MRI use, simply
hypothesizing a role for this technique, especially in high-risk disease. A recent meta-analysis [3] concluded that MRI has a good specificity for T staging, while sensitivity is highly variable. As such, there remains a high-level of clinical uncertainty regarding the potential role of MRI, particularly when compared with conventional, clinically-based risk classification models. In the present study, we compared MRI with existing models (the Cancer of the Prostate Risk Assessment [CAPRA] score and the
http://dx.doi.org/10.1016/j.eururo.2016.08.015 0302-2838/# 2016 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Please cite this article in press as: Morlacco A, et al. The Incremental Role of Magnetic Resonance Imaging for Prostate Cancer Staging before Radical Prostatectomy. Eur Urol (2016), http://dx.doi.org/10.1016/j.eururo.2016.08.015
EURURO-6970; No. of Pages 4 2
EUROPEAN UROLOGY XXX (2016) XXX–XXX
Partin tables) [4,5] and analyzed if the addition of MRI information could improve these clinical models in terms of pathological outcomes at the time of radical prostatectomy (RP). All patients who underwent preoperative 1.5-Tesla MRI and subsequent RP with pelvic lymph node dissection at Mayo Clinic Rochester, MN, USA, between 2003 and 2013 were included in this retrospective, Institutional Review Board approved study. Detailed methods are reported in Supplementary Data 1. In brief, we used age, prostate-specific antigen (PSA) at diagnosis, clinical stage (digital rectal exam), primary and secondary Gleason grade, Gleason score, percentage of positive/total cores to calculate Partin Table estimates and CAPRA scores. MRI results for ECE, SVI, and nodal disease (N+) were collected. Pathological features (pT, pN, Gleason score, surgical margin status, SVI, and ECE) were recorded. We developed logistic
multivariable regression models including: clinical variables (PSA, clinical stage, percentage of involved cores/total cores, primary Gleason 4–5), Partin Table estimates (for each specific outcome) and CAPRA scores; MRI results (in terms of negative/positive exam for each outcome) were then added in each model and the multivariate modeling was reassessed. The predictive ability of each model was further compared developing receiver-operating characteristic curves and analyzing the area under the curve (AUC) before and after the addition of MRI. Five hundred and one patients were included in the final analysis. Demographic, clinical, and pathological features are shown in Supplementary Table 1. ECE, SVI, and N+ were present in 42.3%, 30.7%, and 16.0% of patients, while MRI results were positive for ECE in 147 patients (29.3%), for SVI in 83 patients (16.6%), and for N+ disease in 24 (4.8%). As shown in Table 1, the multivariate clinical models for
Table 1 – Multivariate modeling analysis for predictors of histological outcome ECE prediction
Without MRI OR
Clinical variables PSA % core involv. Age at surgery Clinical stage Primary Gleason 4/5 MRI ECE result Partin ECE MRI ECE result CAPRA score MRI ECE result
1.025 1.014 1.024 1.348 2.344 — 1.042 — 1.328 —
95% CI
1.000 1.005 0.990 1.124 1.374 — 1.019 — 1.191 —
SVI prediction
1.007 1.026 0.997 1.273 5.707 — 1.109 — 1.507 —
95% CI
0.983 1.015 0.959 1.048 2.820 — 1.075 — 1.332 —
N+ prediction
OR
0.054 0.003 0.165 0.001 0.002 — 0.0003 — <0.0001 —
1.021 1.015 1.016 1.220 4.145 2.379 1.023 7.522 1.244 5.248
95% CI
0.991 1.004 0.976 0.975 2.194 1.262 0.995 4.108 1.095 2.994
1.033 1.037 1.037 1.547 11.552 — 1.145 — 1.706 —
1.023 1.024 1.003 1.304 7.147 — 1.115 — 1.660 —
95% CI
0.996 1.009 0.954 1.025 2.066 — 1.074 — 1.415 —
1.051 1.039 1.054 1.658 24.726 — 1.157 — 1.948 —
p value
1.052 1.026 1.058 1.525 7.833 4.488 1.052 13.773 1.413 9.198
0.173 0.007 0.427 0.081 <0.0001 0.007 0.102 <.0001 0.0008 <0.0001
With MRI p value
OR
0.5572 <0.0001 0.8838 0.0148 <0.0001 — <0.0001 — <0.0001 —
1.010 1.027 0.984 1.152 5.073 7.024 1.098 7.582 1.408 8.854
95% CI
0.983 1.015 0.940 0.922 2.233 3.003 1.058 3.449 1.222 4.286
Without MRI OR
Clinical variables PSA % involv core Age at surgery Clinical stage Primary Gleason 4/5 MRI nodes result Partin nodes MRI nodes CAPRA score MRI nodes
p value
Without MRI OR
Clinical variables PSA % core involv. Age at surgery Clinical stage Primary Gleason 4/5 MRI SVI result Partin SVI MRI SVI result CAPRA score MRI SVI result
1.051 1.023 1.060 1.616 4.000 — 1.065 — 1.481 —
With MRI
p value
1.038 1.040 1.029 1.441 11.524 16.431 1.139 16.669 1.622 18.288
0.4721 <0.0001 0.4768 0.2130 0.0001 <.0001 <0.0001 <0.0001 <0.0001 <0.0001
With MRI p value
OR
0.0964 0.0012 0.9170 0.0304 0.0019 — <0.0001 — <0.0001 —
1.032 1.020 0.986 1.344 9.669 4.831 1.102 14.249 1.682 15.060
95% CI
0.996 1.003 0.928 1.014 2.263 1.320 1.056 3.141 1.392 4.348
p value
1.068 1.037 1.048 1.782 41.309 17.674 1.150 64.638 2.032 52.165
0.0794 0.0235 0.6558 0.0395 0.0022 0.0173 <0.0001 0.0006 <0.0001 <0.001
CAPRA = Cancer of the Prostate Risk Assessment; CI = confidence interval; ECE = extracapsular extension; MRI = magnetic resonance imaging; N+ = lymph-node invasion; OR = odds ratio; PSA = prostate-specific antigen; SVI = seminal vesicle invasion.
Please cite this article in press as: Morlacco A, et al. The Incremental Role of Magnetic Resonance Imaging for Prostate Cancer Staging before Radical Prostatectomy. Eur Urol (2016), http://dx.doi.org/10.1016/j.eururo.2016.08.015
EURURO-6970; No. of Pages 4 EUROPEAN UROLOGY XXX (2016) XXX–XXX
3
Fig. 1 – receiver operating characteristic (ROC) curves for extracapsular extension (ECE) and seminal vesicle invasion (SVI) prediction. CAPRA = Cancer of the Prostate Risk Assessment; MRI = magnetic resonance imaging.
prediction of ECE, SVI, and nodal disease included age, PSA, clinical stage, primary Gleason score 4 or 5, and percentage of involved cores/total cores. All of these, except for age and PSA, were significant predictors of outcome, with primary Gleason 4 or 5 being the strongest factor associated with each outcome (odds ratio [OR] 2.3, p < 0.01 for ECE; OR 5.7, p < 0.0001 for SVI; OR 7.147, p < 0.01 for N+ disease). When MRI results were added to each model, they showed a significant association with the outcomes (all p < 0.001). Combined MRI and each model outperformed individual models alone at all levels in the receiver-operating characteristic analysis (Supplementary Table 2). Comparing Partin and Partin + MRI predictive models, the AUC was 0.61 versus 0.73 for ECE, 0.75 versus 0.82 for SVI (Fig. 1), and 0.82 versus 0.85 for N+ (Supplementary Fig. 1). Comparing CAPRA and CAPRA + MRI models, the AUC was 0.69 versus 0.77 for ECE, 0.75 versus 0.83 for SVI (Fig. 1), 0.82 versus 0.85 for N+ (Supplementary Fig. 1). Our results suggest that MRI information can provide an added value when compared with clinical-based, Partin Table, or CAPRA score models alone for the prediction of adverse histopathological outcomes at RP. Based on our data and on the nature of our cohort, patients with doubtful clinical findings and intermediate to high risk disease could
benefit the most from preoperative MRI. Few studies have explored the potential additional value of MRI, and those that have are relatively underpowered. Feng et al [6] compared the predictive accuracy of MRI and clinical models (Partin Tables and Memorial Sloan Kettering Cancer Center nomogram) for ECE finding a minor improvement in diagnostic accuracy after addition of MRI to the model (AUC for Partin Tables and Memorial Sloan Kettering Cancer Center of 0.85 and 0.86, respectively, increased to 0.93 and 0.94 after addition of MRI). The magnitude of the difference they detected is likely lower than ours because they had a lower risk cohort and thus had lower rates of ECE, which would falsely inflate the AUC for models that predicted against ECE and underestimate the utility of an MRI. For instance, 95% of their cohort was either cT1c or cT2a, while only 66% of our core met these criteria. Gupta et al [7] compared the accuracy of 3-Tesla MRI and Partin Tables in predicting ECE. This study did not analyze the additional value of MRI to the Partin Tables, but rather compared those two tools in alternative models, finding a Partin accuracy similar to ours (AUC: 0.62) and a MRI AUC of 0.82. Again, however, the study population was small (60 patients) and with a clear preponderance of clinically low and intermediate-risk patients. Tay et al [8] used 3-Tesla
Please cite this article in press as: Morlacco A, et al. The Incremental Role of Magnetic Resonance Imaging for Prostate Cancer Staging before Radical Prostatectomy. Eur Urol (2016), http://dx.doi.org/10.1016/j.eururo.2016.08.015
EURURO-6970; No. of Pages 4 4
EUROPEAN UROLOGY XXX (2016) XXX–XXX
MRI for ECE prediction, finding an AUC of 0.69 for clinical model alone, 0.72 for clinical model + nonspecializedreading MRI and 0.91 for the two previous elements + specialized (dedicated radiologist) MRI reading. In our study, all the radiologists involved in final MRI readings were consultants with a consolidated experience in genitourinary pathology, and the incremental benefit was already evident at first reading. Our study has several limitations, including the retrospective nature, the high percentage of high-risk patients in the cohort, the long study period, and the lack of use of the Prostate Imaging Reporting and Data System system [9]. A detailed discussion of these aspects can be found in Supplementary Data 2. In conclusion, MRI provides additional predictive value to clinical-based models alone, but it is possible that our findings underestimate the true diagnostic accuracy of MRI and its added value. As such, prospective investigations with Prostate Imaging Reporting and Data System and multiparametric MRI use are warranted to clarify the ability of MRI to improve clinical prediction in this setting. Author contributions: R. Jeffrey Karnes had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Morlacco, Karnes, Sharma, Froemming. Acquisition of data: Morlacco, Carlson, Rangel, Sharma, Viers. Analysis and interpretation of data: Morlacco, Rangel, Carlson, Froemming, Karnes. Drafting of the manuscript: Morlacco, Karnes, Viers, Sharma, Froemming.
Funding/Support and role of the sponsor: This publication was made possible by Clinical and Translational Science Award Grant Number UL1 TR000135 from the National Center for Advancing Translational Sciences, a component of the National Institutes of Health. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of National Institutes of Health.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. eururo.2016.08.015. References [1] EAU 2015 Guidelines on Prostate Cancer. https://uroweb.org/ guideline/prostate-cancer/. Accessed November 28, 2015. [2] NCCN. NCCN Guidelines on Prostate Cancer.1.2016. https://www. nccn.org/store/login/login.aspx?ReturnURL=http://www.nccn.org/ professionals/physician_gls/pdf/prostate.pdf. [3] de Rooij M, Hamoen EHJ, 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. [4] Cooperberg MR, Pasta DJ, Elkin EP, et al. The University of California, San Francisco Cancer of the Prostate Risk Assessment score: A straightforward and reliable preoperative predictor of disease recurrence after radical prostatectomy. J Urol 2005;173:1938–42. [5] Eifler JB, Feng Z, Lin BM, et al. An updated prostate cancer staging nomogram (Partin tables) based on cases from 2006 to 2011. BJU Int 2013;111:22–9. [6] Feng TS, Sharif-Afshar AR, Wu J, et al. Multiparametric MRI improves
Critical revision of the manuscript for important intellectual content:
accuracy of clinical nomograms for predicting extracapsular exten-
Morlacco, Sharma, Viers, Rangel, Carlson, Froemming, Karnes.
sion of prostate cancer. Urology 2015;86:332–7.
Statistical analysis: Rangel, Carlson, Morlacco.
[7] Gupta RT, Faridi KF, Singh AA, et al. Comparing 3-T multiparametric
Obtaining funding: None.
MRI and the Partin tables to predict organ-confined prostate cancer
Administrative, technical, or material support: None.
after radical prostatectomy. Urol Oncol 2014;32:1292–9.
Supervision: Karnes, Froemming. Other: None.
[8] Tay KJ, Gupta RT, Brown AF, Silverman RK, Polascik TJ. Defining the incremental utility of prostate multiparametric magnetic resonance imaging at standard and specialized read in predicting extracapsular
Financial disclosures: R. Jeffrey Karnes certifies that all conflicts of interest, including specific financial interests and relationships and
extension of prostate cancer. Eur Urol 2016;70:211–3. [9] Radtke JP, Hadaschik B, Wolf M, et al. The impact of magnetic
affiliations relevant to the subject matter or materials discussed in the
resonance imaging on prediction of extraprostatic extension and
manuscript (eg, employment/affiliation, grants or funding, consultan-
prostatectomy outcome in low-, intermediate- and high-risk pros-
cies, honoraria, stock ownership or options, expert testimony, royalties,
tate cancer patients. Try to find a standard. J Endourol 2015;29:
or patents filed, received, or pending), are the following: None.
1396–405.
Please cite this article in press as: Morlacco A, et al. The Incremental Role of Magnetic Resonance Imaging for Prostate Cancer Staging before Radical Prostatectomy. Eur Urol (2016), http://dx.doi.org/10.1016/j.eururo.2016.08.015