Oncology Added Value of Multiparametric Ultrasonography in Magnetic Resonance Imaging and Ultrasonography Fusioneguided Biopsy of the Prostate in Patients With Suspicion for Prostate Cancer Andreas Maxeiner, Carsten Stephan, Tahir Durmus, Torsten Slowinski, Hannes Cash, and Thomas Fischer OBJECTIVE
METHODS
RESULTS
CONCLUSION
To analyze whether magnetic resonance imaging-ultrasonography (MRI-US) fusioneguided biopsy detects more and clinical significant prostate cancer (PCa) in comparison to conventional transrectal USeguided prostate biopsy (PBX) and to investigate if multiparametric (mp) US during MRI-US fusion can further characterize mpMRI-suspected lesions according to the prostate MRI reporting and data system (PI-RADS). From January 2012 to January 2014, 169 patients with a median of 2 negative conventional PBX and/or initially or consistently elevated prostate-specific antigen levels were prospectively included and underwent 3 T mpMRI. Real-time MRI-US fusion scan was used to biopsy the mpMRI-targeted lesions (n ¼ 316). Scanning by mpUS, including B-mode, power Doppler, strain elastography, and contrast-enhanced US was performed to further characterize those lesions and to score by US modalities resulting in an mpUS score. Afterward, a conventional 10-core PBX was performed. PCa detection based on the results of targeted and conventional PBX was estimated. Performances of single US modalities were analyzed. The mpUS score was also investigated for PCa and PI-RADS score prediction. Among 169 patients, 71 PCa (42%) were detected. From these 71 cases, clinically significant PCa (Gleason score 7) were detected exclusively by MRI-US fusion in 31 from 46 cases (67.4%). The highest sensitivity was observed in contrast-enhanced US (85%) and elastography (80%). The mpUS score predicts PCa and PI-RADS score with an overall accuracy of 86% and 80%, respectively. MRI-US fusioneguided PBX detects more clinically significant PCa compared with conventional TRUS. The mpUS score correlates with PI-RADS in PCa prediction. UROLOGY 86: 108e114, 2015. 2015 Elsevier Inc.
P
rostate cancer (PCa) represents the most frequently diagnosed malignant disease among adult males in the United States of America and the second leading cause of cancer-related deaths in men.1 The increasing accuracy of multiparametric
Financial Disclosure: The authors declare that they have no relevant financial interests. From the Department of Urology, Charite-Universitaetsmedizin Berlin, Berlin, Germany; the Berlin Institute for Urologic Research, Berlin, Germany; and the Department of Radiology, Charite-Universitaetsmedizin Berlin, Berlin, Germany; and the Department of Nephrology, Charite-Universitaetsmedizin Berlin, Berlin, Germany. Address correspondence to: Andreas Maxeiner, M.D., Department of Urology, Charite-Universitaetsmedizin Berlin, Chariteplatz 1, Berlin 10117, Germany. E-mail: [email protected] Submitted: September 16, 2014, accepted (with revisions): January 29, 2015
108
ª 2015 Elsevier Inc. All Rights Reserved
magnetic resonance imaging (mpMRI) has greatly improved the ability of localizing PCa foci and seems to become an integral part of PCa diagnosis.2 The mpMRI protocol usually consists of morphological T2-weighted (T2w) sequences with functional imaging, that is, diffusion-weighted imaging, dynamic contrast-enhanced imaging, and sometimes MR spectroscopy.3 Because of decreased detection rates in repeated conventional transrectal ultrasonography (TRUS)-guided prostate biopsy, more men with prior negative biopsies and persistently elevated prostate-specific antigen (PSA) levels undergo mpMRI of the prostate. A structured prostate imaging reporting and data system (PI-RADS) was published in 2012 to stratify the likelihood of PCa in http://dx.doi.org/10.1016/j.urology.2015.01.055 0090-4295/15
UROLOGY 86 (1), 2015
109
prostatic lesions to improve inter-reader reproducibility.4,5 Lesions suspicious for PCa are classified by PI-RADS from “clinically significant disease is highly unlikely to be present” (score 1) to “clinically significant cancer is highly likely to be present” (score 5).5 Real-time MRI-US fusioneguided prostate biopsy aligns an MR image with a transrectal US image to target a prostate biopsy toward a suspected lesion. Thus, the overdiagnosis of insignificant tumors may be reduced, sparing the patient an unnecessary radical treatment.6 The detection of significant PCa (determined as cT2b and/or Gleason grade 7) is increased by using MRI-US fusioneguided prostate biopsy in patients with prior negative biopsies.7,8 Because PCa is not visible in approximately 50% of conventional gray-scale ultrasonography, other ultrasonography modalities such as color and power Doppler, contrast mediaeenhanced US (CEUS), or strain elastography or tissue Doppler imaging (TDI), respectively, have been developed.9-11 To our knowledge, a combination of the mentioned ultrasonography modalities, referred in our study as multiparametric ultrasonography (mpUS), has not been evaluated in general or in MRI-US fusioneguided prostate biopsy. Furthermore, the literature describes several methods of MRI-US image registration: (1) cognitive registration, (2) sensor-based registration, and (3) organ-based registration.8,12,13 In our study, mpMRI was fused to a TRUS image via a softwarebased image or sensor-based coregistration. The aim of this prospective study was (1) to compare detection rates of real-time MRI-US fusioneguided and conventional TRUS-guided prostate biopsy especially for the detection of significant PCa; (2) to evaluate mpUS and a calculated score for classification of mpMRItargeted lesion during real-time MRI-US fusion biopsy to predict PCa and PI-RADS score.
METHODS Study Design and Population During the study period spanning from January 2012 to January 2014, 169 consecutive men with persistent or initial suspicion for PCa (elevated PSA levels or suspicious digital rectal examination) and with a median of 2 negative biopsies of the prostate were included; in addition, 18 cases with initial biopsy were incorporated in this study. All patients gave written informed consent and underwent prebiopsy mpMRI. The mpMRI findings were classified by a single radiologist based on the PI-RADS score. The mpMR images were fused to a TRUS image. The MR suspicious lesions (n ¼ 316) were also scored based on mpUS modalities by another radiologist (aware of the MR predefined location but not
of the PI-RADS score) and finally biopsied by a distinguished urologist. The subsequent conventional 10-core biopsy was also performed by the same urologist. All patients received an antibiotic prophylactic treatment before the biopsy.
Multiparametric MRI and PI-RADS All patients underwent mpMRI before the MRI-US fusion biopsy on commercially available 3 T MRI (Skyra; Siemens, Erlangen, Germany) using 18-channel surface coils. As previously reported,14 the MRI protocol included T2-weighted turbo spin-echo sequences in an axial and a coronal slice orientation (repetition time (TR)/echo time (TE), 4000/101 ms; echo train length, 25; averages 2; field of view (FOV), 180 180 mm; thickness, 3 mm; distance factor, 10%; image matrix, 320 320; phase oversampling, 100%), an axial T1-weighted turbo spin-echo sequence (TR/TE, 700/12 ms; echo train length, 3; averages 2; FOV, 200 200 mm; thickness, 3 mm; distance factor, 20%, image matrix, 256 192; phase oversampling, 80%), and further diffusion-weighted images with apparent diffusion coefficient (ADC) map calculation (TR/TE, 3500/ 80 ms; averaging 8; FOV, 200 200 mm; image matrix, 128 128; thickness, 3 mm; b values: 0, 100, 500, 1000 s/mm). Every lesion suspicious for a tumor within the prostate was assigned a point value between 1 and 5 based on clearly defined criteria according to PI-RADS. The highest PI-RADS score of a particular suspicious lesion was included for further analysis.
MRI-US Fusion Two similar commercial ultrasonography workstations and probes, (1) Aplio 500 (Toshiba, Otawara, Japan; with an endfire 11C3 probe) for the fusion-guided biopsy and (2) HI VISION Preirus (Hitachi Medical Systems GmbH, Wiesbaden, Germany; with a biplane EUP CC531 probe) for conventional TRUS, were used. The patient was placed in a left lateral decubitus position. The B-mode frequency was adjusted depending on the depth of penetration between 7.5 and 11 MHz. After transferring the Digital Imaging and Communications in Medicine (DICOM) MRI data set to the US system, a software-based image and sensor-based coregistration was used. The probe was determined by a tracking generator (miniBird receiver [Ascension Tech Corp., Burlington, VT], implemented on the US platform). A low magnetic field was administered (0.1 T) so that the position sensor to the probe base was localized in space. The magnetic field was directly installed in front of the patient’s pelvis. The movement of the probe could be traced within the magnetic field space and MRI and US images could be aligned simultaneously in all spatial directions.
Multiparametric US and MRI-US Fusioneguided Prostate Biopsy According to the PI-RAD system, we scored semiquantitatively using an analog scale from 1 (clinically significant disease is
=Figure 1. Case A (age 69 years; PSA [prostate-specific antigen level] ¼ 9.01 ng/mL; Gleason 4 þ 4 ¼ 8; highest prostate imaging reporting and data system [PI-RADS] score ¼ 5; multiparametric ultrasonography score [mpUS] ¼ 15). Case B (age 53 years; PSA ¼ 5.94 ng/mL; no evidence of malignancy in the biopsy result; highest PI-RADS score ¼ 3; mpUS score ¼ 5). Ultrasonography modalities: (A,B) B-mode; (C,D) power Doppler; (E,F) strain elastography or tissue Doppler imaging; (G,H) contrast-enhanced ultrasonography; and (I,J) final magnetic resonance imaging-ultrasonography fusioneguided biopsy of each lesion; the appearance of the T2w magnetic resonance images during initial fusion is always included. (Color version available online.)
110
UROLOGY 86 (1), 2015
UROLOGY 86 (1), 2015
(62.0) (52.0) (42.9) (85.7) (92.9) (75.0) 44* 13 9 6 13 3 (100) (35.2) (29.6) (9.86) (19.7) (5.64)
N (%)
71 25 21 7 14 4 PCa cases (71/169), N (%) Gleason score ¼ 6, N (%) Gleason score [ 7 (3 D 4), N (%) Gleason score [ 7 (4 D 3), N (%) Gleason score [ 8, N (%) Gleason score [ 9, N (%)
The mean age of the patient population was 65.6 years (7.71 years), the mean prebiopsy PSA level was 13.9 ng/mL (13.69 ng/mL), and the mean estimated prostate volume was 60.6 cm3 (35.3 cm3). The patients enrolled in our study had a median of 2 (range 0-9) prior negative biopsies. A suspicious digital rectal examination was observed in 12 subjects. For the mpMRI suspicious lesions, the highest PI-RADS score was used for analysis, whereas a median PI-RADS score of 3 (range 2-5) was assigned; PI-RADS 4 (n ¼ 38) and 5 (n ¼ 39) patients
Gleason Score
RESULTS
Table 1. Distribution of biopsy Gleason scores
Data analysis was performed using SPSS version 15.0 software (IBM Corp., Armonk, NY): Categorical variable comparisons were performed using the chi-square or Fisher exact test, and continuous variables were evaluated with the Student t test or Mann-Whitney test, as appropriate. A P value < .05 was used to define significant results. Clinically significant PCa was defined as Gleason score 7. For evaluation of the predictive power of the mpUS score, receiver operating characteristic analysis was used. Optimal cutoff values were taken from corresponding receiver operating characteristic curves. Sensitivities, specificities, and positive and negative predictive values were calculated out of SPSS processed cross-tables.
MRI-US Fusion
Statistical Analysis
mpUS, multiparametric ultrasonography; MRI, magnetic resonance imaging; PCa, prostate cancer; PI-RADS, prostate imaging reporting and data system; TRUS, transrectal ultrasonography; US, ultrasonography. Biopsy Gleason scores observed on conventional TRUS-guided 10-core biopsy vs MRI-US fusioneguided biopsy and PCa detection rates by the highest average PI-RADS score per lesion and mpUS score (applied on the cutoff: 8.5). Values in bold indicate clinically significant cancer. * Forty of 44 PCa cases were detected in both techniques. Because the same urologist performed the biopsy also conventionally, the mpMRI-suspected lesion was biopsied twice in almost every case. y Twenty-seven PCa cases have been solely detected by TRUS-guided biopsy. The percentages are applied horizontally, except for the first column (N [%]).
1 (7.20)
19 (27.8) 8 (32.0) 10 (47.6)
(73.2) (68.0) (52.4) (100) (92.8) (100) 52 17 11 7 13 4 (46.5) (12.0) (47.6) (57.1) (85.7) (100) 33 3 10 4 12 4 (32.4) (40.0) (42.9) (28.6) (14.3) 23 10 9 2 2 (14.1) (28.0) (9.50) (14.3) 10 7 2 1 5 (7.00) 5 (20.0) (38.0) (48.0) (57.1) (14.3) (7.10) (25.0) 27 12 12 1 1 1
mpUS 8.5 Highest PI-RADS 5 Highest PI-RADS 4 Highest PI-RADS 3 Highest PI-RADS 2 TRUS
y
highly unlikely to be present) to 2 (clinically significant cancer is equivocal) to 3 (clinically significant cancer is highly likely to be present) for every single US modality. On the basis of differences in image acquisition between US (oblique plane) and MRI (axial plane), an angulation of 30 during initial fusion needs to be accounted, resulting in a modified appearance of the T2w MR images. These modified T2w MR images were scored similarly to the subsequent mpUS scoring (3 ¼ confluent hypointense area; 2 ¼ inhomogeneous area with mild hypointensity; 1 ¼ uniform hyperintense T2 signal). Thus, in combination with the mpUS including 4 US modalities (each 1-3 points), a total of maximum 15 points was scored. The mpUS modalities are illustrated by 2 cases within Figure 1: Bmode (Fig. 1A,B), power Doppler (Fig. 1C,D), strain elastography in combination with TDI (Fig. 1E,F), respectively and CEUS (Fig. 1G,H; contrast software with a low MI of 0.12; 2.4 mL of SonoVue intravenous contrast medium by Bracco, Milan, Italy, and further 5 mL sodium chloride 0.9% intravenous). Because there is no reported mpUS scoring, a detailed explanation for all methods will be given as follows: B-mode: 3 ¼ hypoechoic parenchyma, 2 ¼ inconclusive, 1 ¼ isoechoic parenchyma; power Doppler (vascularity): 3 ¼ hypervascular (high), 2 ¼ indifferent with atypical pattern (moderate), 1 ¼ hypovascular (low); strain elastography in combination with TDI: 3 ¼ dark blue colored (hard) and a fully extinguished area in the TDI, 2 ¼ blue or green and a partially extinguished area in the TDI, 1 ¼ green or red (soft) without an extinguished area in the TDI; CEUS: 3 ¼ early and strong enhancement and early washout, 2 ¼ early enhancement but no washout, 1 ¼ isoechogenic enhancement. As a proof of concept, the US modalities were scored fairly subjectively. After mpUS and clear reidentification of anatomic landmarks (seminal vesicles and Denonvilliers fascia), an MRI-US fusion biopsy was performed (Fig. 1I,J), followed by a conventional TRUS-guided 10-core biopsy without MRI-US fusion.
111
Table 2. Ultrasonography modalityebased performances Ultrasonography Modalities B-mode Power Doppler Strain elastography/tissue Doppler imaging CEUS T2w MRI appearance during initial US fusion
Sensitivity
Specificity
PPV
NPV
0.73 0.63 0.8 0.85 0.92
0.67 0.62 0.63 0.8 0.49
0.62 0.55 0.61 0.75 0.57
0.78 0.7 0.82 0.88 0.89
Mean Score Standard Deviation 1.63 1.57 1.68 1.62 1.85
0.73 0.66 0.7 0.74 0.7
CEUS, contrast-enhanced ultrasonography; NPV, negative predictive value; positive predictive value; T2w, T2 weighted. Other abbreviations as in Table 1. Calculated sensitivities, specificities, PPV, and NPV for PCa prediction of single ultrasonography modalities applied on PI-RADS predefined lesions.
comprised 45.6% of the total cohort. A total of 316 targeted lesions were biopsied. The mean number of targeted biopsies per patient was 1.86 (0.55). The mean total number of biopsies including the subsequent 10 cores (total n ¼ 1690) was 11.81 (0.82) per patient. The lesion-based analysis showed a cancer detection rate for the target biopsy of 21.2% (67 of 316) and 11.8% (200 of 1690) for the random biopsy. The overall cancer detection rate was 42% (71 of 169 patients). From 71 PCa patients, 48 (67.6%) underwent radical prostatectomy, where 5 patients (7.05%) underwent radiation therapy, 3 patients (4.23%) received hormonal therapy, and 15 (21.1%) were included in an active surveillance program. In accordance with the Standards of Reporting for MRItargeted Biopsy Studies working group recommendations,15 we divided our cohort into a clinically insignificant group (Gleason score
METHODS
RESULTS
CONCLUSION
To analyze whether magnetic resonance imaging-ultrasonography (MRI-US) fusioneguided biopsy detects more and clinical significant prostate cancer (PCa) in comparison to conventional transrectal USeguided prostate biopsy (PBX) and to investigate if multiparametric (mp) US during MRI-US fusion can further characterize mpMRI-suspected lesions according to the prostate MRI reporting and data system (PI-RADS). From January 2012 to January 2014, 169 patients with a median of 2 negative conventional PBX and/or initially or consistently elevated prostate-specific antigen levels were prospectively included and underwent 3 T mpMRI. Real-time MRI-US fusion scan was used to biopsy the mpMRI-targeted lesions (n ¼ 316). Scanning by mpUS, including B-mode, power Doppler, strain elastography, and contrast-enhanced US was performed to further characterize those lesions and to score by US modalities resulting in an mpUS score. Afterward, a conventional 10-core PBX was performed. PCa detection based on the results of targeted and conventional PBX was estimated. Performances of single US modalities were analyzed. The mpUS score was also investigated for PCa and PI-RADS score prediction. Among 169 patients, 71 PCa (42%) were detected. From these 71 cases, clinically significant PCa (Gleason score 7) were detected exclusively by MRI-US fusion in 31 from 46 cases (67.4%). The highest sensitivity was observed in contrast-enhanced US (85%) and elastography (80%). The mpUS score predicts PCa and PI-RADS score with an overall accuracy of 86% and 80%, respectively. MRI-US fusioneguided PBX detects more clinically significant PCa compared with conventional TRUS. The mpUS score correlates with PI-RADS in PCa prediction. UROLOGY 86: 108e114, 2015. 2015 Elsevier Inc.
P
rostate cancer (PCa) represents the most frequently diagnosed malignant disease among adult males in the United States of America and the second leading cause of cancer-related deaths in men.1 The increasing accuracy of multiparametric
Financial Disclosure: The authors declare that they have no relevant financial interests. From the Department of Urology, Charite-Universitaetsmedizin Berlin, Berlin, Germany; the Berlin Institute for Urologic Research, Berlin, Germany; and the Department of Radiology, Charite-Universitaetsmedizin Berlin, Berlin, Germany; and the Department of Nephrology, Charite-Universitaetsmedizin Berlin, Berlin, Germany. Address correspondence to: Andreas Maxeiner, M.D., Department of Urology, Charite-Universitaetsmedizin Berlin, Chariteplatz 1, Berlin 10117, Germany. E-mail: [email protected] Submitted: September 16, 2014, accepted (with revisions): January 29, 2015
108
ª 2015 Elsevier Inc. All Rights Reserved
magnetic resonance imaging (mpMRI) has greatly improved the ability of localizing PCa foci and seems to become an integral part of PCa diagnosis.2 The mpMRI protocol usually consists of morphological T2-weighted (T2w) sequences with functional imaging, that is, diffusion-weighted imaging, dynamic contrast-enhanced imaging, and sometimes MR spectroscopy.3 Because of decreased detection rates in repeated conventional transrectal ultrasonography (TRUS)-guided prostate biopsy, more men with prior negative biopsies and persistently elevated prostate-specific antigen (PSA) levels undergo mpMRI of the prostate. A structured prostate imaging reporting and data system (PI-RADS) was published in 2012 to stratify the likelihood of PCa in http://dx.doi.org/10.1016/j.urology.2015.01.055 0090-4295/15
UROLOGY 86 (1), 2015
109
prostatic lesions to improve inter-reader reproducibility.4,5 Lesions suspicious for PCa are classified by PI-RADS from “clinically significant disease is highly unlikely to be present” (score 1) to “clinically significant cancer is highly likely to be present” (score 5).5 Real-time MRI-US fusioneguided prostate biopsy aligns an MR image with a transrectal US image to target a prostate biopsy toward a suspected lesion. Thus, the overdiagnosis of insignificant tumors may be reduced, sparing the patient an unnecessary radical treatment.6 The detection of significant PCa (determined as cT2b and/or Gleason grade 7) is increased by using MRI-US fusioneguided prostate biopsy in patients with prior negative biopsies.7,8 Because PCa is not visible in approximately 50% of conventional gray-scale ultrasonography, other ultrasonography modalities such as color and power Doppler, contrast mediaeenhanced US (CEUS), or strain elastography or tissue Doppler imaging (TDI), respectively, have been developed.9-11 To our knowledge, a combination of the mentioned ultrasonography modalities, referred in our study as multiparametric ultrasonography (mpUS), has not been evaluated in general or in MRI-US fusioneguided prostate biopsy. Furthermore, the literature describes several methods of MRI-US image registration: (1) cognitive registration, (2) sensor-based registration, and (3) organ-based registration.8,12,13 In our study, mpMRI was fused to a TRUS image via a softwarebased image or sensor-based coregistration. The aim of this prospective study was (1) to compare detection rates of real-time MRI-US fusioneguided and conventional TRUS-guided prostate biopsy especially for the detection of significant PCa; (2) to evaluate mpUS and a calculated score for classification of mpMRItargeted lesion during real-time MRI-US fusion biopsy to predict PCa and PI-RADS score.
METHODS Study Design and Population During the study period spanning from January 2012 to January 2014, 169 consecutive men with persistent or initial suspicion for PCa (elevated PSA levels or suspicious digital rectal examination) and with a median of 2 negative biopsies of the prostate were included; in addition, 18 cases with initial biopsy were incorporated in this study. All patients gave written informed consent and underwent prebiopsy mpMRI. The mpMRI findings were classified by a single radiologist based on the PI-RADS score. The mpMR images were fused to a TRUS image. The MR suspicious lesions (n ¼ 316) were also scored based on mpUS modalities by another radiologist (aware of the MR predefined location but not
of the PI-RADS score) and finally biopsied by a distinguished urologist. The subsequent conventional 10-core biopsy was also performed by the same urologist. All patients received an antibiotic prophylactic treatment before the biopsy.
Multiparametric MRI and PI-RADS All patients underwent mpMRI before the MRI-US fusion biopsy on commercially available 3 T MRI (Skyra; Siemens, Erlangen, Germany) using 18-channel surface coils. As previously reported,14 the MRI protocol included T2-weighted turbo spin-echo sequences in an axial and a coronal slice orientation (repetition time (TR)/echo time (TE), 4000/101 ms; echo train length, 25; averages 2; field of view (FOV), 180 180 mm; thickness, 3 mm; distance factor, 10%; image matrix, 320 320; phase oversampling, 100%), an axial T1-weighted turbo spin-echo sequence (TR/TE, 700/12 ms; echo train length, 3; averages 2; FOV, 200 200 mm; thickness, 3 mm; distance factor, 20%, image matrix, 256 192; phase oversampling, 80%), and further diffusion-weighted images with apparent diffusion coefficient (ADC) map calculation (TR/TE, 3500/ 80 ms; averaging 8; FOV, 200 200 mm; image matrix, 128 128; thickness, 3 mm; b values: 0, 100, 500, 1000 s/mm). Every lesion suspicious for a tumor within the prostate was assigned a point value between 1 and 5 based on clearly defined criteria according to PI-RADS. The highest PI-RADS score of a particular suspicious lesion was included for further analysis.
MRI-US Fusion Two similar commercial ultrasonography workstations and probes, (1) Aplio 500 (Toshiba, Otawara, Japan; with an endfire 11C3 probe) for the fusion-guided biopsy and (2) HI VISION Preirus (Hitachi Medical Systems GmbH, Wiesbaden, Germany; with a biplane EUP CC531 probe) for conventional TRUS, were used. The patient was placed in a left lateral decubitus position. The B-mode frequency was adjusted depending on the depth of penetration between 7.5 and 11 MHz. After transferring the Digital Imaging and Communications in Medicine (DICOM) MRI data set to the US system, a software-based image and sensor-based coregistration was used. The probe was determined by a tracking generator (miniBird receiver [Ascension Tech Corp., Burlington, VT], implemented on the US platform). A low magnetic field was administered (0.1 T) so that the position sensor to the probe base was localized in space. The magnetic field was directly installed in front of the patient’s pelvis. The movement of the probe could be traced within the magnetic field space and MRI and US images could be aligned simultaneously in all spatial directions.
Multiparametric US and MRI-US Fusioneguided Prostate Biopsy According to the PI-RAD system, we scored semiquantitatively using an analog scale from 1 (clinically significant disease is
=Figure 1. Case A (age 69 years; PSA [prostate-specific antigen level] ¼ 9.01 ng/mL; Gleason 4 þ 4 ¼ 8; highest prostate imaging reporting and data system [PI-RADS] score ¼ 5; multiparametric ultrasonography score [mpUS] ¼ 15). Case B (age 53 years; PSA ¼ 5.94 ng/mL; no evidence of malignancy in the biopsy result; highest PI-RADS score ¼ 3; mpUS score ¼ 5). Ultrasonography modalities: (A,B) B-mode; (C,D) power Doppler; (E,F) strain elastography or tissue Doppler imaging; (G,H) contrast-enhanced ultrasonography; and (I,J) final magnetic resonance imaging-ultrasonography fusioneguided biopsy of each lesion; the appearance of the T2w magnetic resonance images during initial fusion is always included. (Color version available online.)
110
UROLOGY 86 (1), 2015
UROLOGY 86 (1), 2015
(62.0) (52.0) (42.9) (85.7) (92.9) (75.0) 44* 13 9 6 13 3 (100) (35.2) (29.6) (9.86) (19.7) (5.64)
N (%)
71 25 21 7 14 4 PCa cases (71/169), N (%) Gleason score ¼ 6, N (%) Gleason score [ 7 (3 D 4), N (%) Gleason score [ 7 (4 D 3), N (%) Gleason score [ 8, N (%) Gleason score [ 9, N (%)
The mean age of the patient population was 65.6 years (7.71 years), the mean prebiopsy PSA level was 13.9 ng/mL (13.69 ng/mL), and the mean estimated prostate volume was 60.6 cm3 (35.3 cm3). The patients enrolled in our study had a median of 2 (range 0-9) prior negative biopsies. A suspicious digital rectal examination was observed in 12 subjects. For the mpMRI suspicious lesions, the highest PI-RADS score was used for analysis, whereas a median PI-RADS score of 3 (range 2-5) was assigned; PI-RADS 4 (n ¼ 38) and 5 (n ¼ 39) patients
Gleason Score
RESULTS
Table 1. Distribution of biopsy Gleason scores
Data analysis was performed using SPSS version 15.0 software (IBM Corp., Armonk, NY): Categorical variable comparisons were performed using the chi-square or Fisher exact test, and continuous variables were evaluated with the Student t test or Mann-Whitney test, as appropriate. A P value < .05 was used to define significant results. Clinically significant PCa was defined as Gleason score 7. For evaluation of the predictive power of the mpUS score, receiver operating characteristic analysis was used. Optimal cutoff values were taken from corresponding receiver operating characteristic curves. Sensitivities, specificities, and positive and negative predictive values were calculated out of SPSS processed cross-tables.
MRI-US Fusion
Statistical Analysis
mpUS, multiparametric ultrasonography; MRI, magnetic resonance imaging; PCa, prostate cancer; PI-RADS, prostate imaging reporting and data system; TRUS, transrectal ultrasonography; US, ultrasonography. Biopsy Gleason scores observed on conventional TRUS-guided 10-core biopsy vs MRI-US fusioneguided biopsy and PCa detection rates by the highest average PI-RADS score per lesion and mpUS score (applied on the cutoff: 8.5). Values in bold indicate clinically significant cancer. * Forty of 44 PCa cases were detected in both techniques. Because the same urologist performed the biopsy also conventionally, the mpMRI-suspected lesion was biopsied twice in almost every case. y Twenty-seven PCa cases have been solely detected by TRUS-guided biopsy. The percentages are applied horizontally, except for the first column (N [%]).
1 (7.20)
19 (27.8) 8 (32.0) 10 (47.6)
(73.2) (68.0) (52.4) (100) (92.8) (100) 52 17 11 7 13 4 (46.5) (12.0) (47.6) (57.1) (85.7) (100) 33 3 10 4 12 4 (32.4) (40.0) (42.9) (28.6) (14.3) 23 10 9 2 2 (14.1) (28.0) (9.50) (14.3) 10 7 2 1 5 (7.00) 5 (20.0) (38.0) (48.0) (57.1) (14.3) (7.10) (25.0) 27 12 12 1 1 1
mpUS 8.5 Highest PI-RADS 5 Highest PI-RADS 4 Highest PI-RADS 3 Highest PI-RADS 2 TRUS
y
highly unlikely to be present) to 2 (clinically significant cancer is equivocal) to 3 (clinically significant cancer is highly likely to be present) for every single US modality. On the basis of differences in image acquisition between US (oblique plane) and MRI (axial plane), an angulation of 30 during initial fusion needs to be accounted, resulting in a modified appearance of the T2w MR images. These modified T2w MR images were scored similarly to the subsequent mpUS scoring (3 ¼ confluent hypointense area; 2 ¼ inhomogeneous area with mild hypointensity; 1 ¼ uniform hyperintense T2 signal). Thus, in combination with the mpUS including 4 US modalities (each 1-3 points), a total of maximum 15 points was scored. The mpUS modalities are illustrated by 2 cases within Figure 1: Bmode (Fig. 1A,B), power Doppler (Fig. 1C,D), strain elastography in combination with TDI (Fig. 1E,F), respectively and CEUS (Fig. 1G,H; contrast software with a low MI of 0.12; 2.4 mL of SonoVue intravenous contrast medium by Bracco, Milan, Italy, and further 5 mL sodium chloride 0.9% intravenous). Because there is no reported mpUS scoring, a detailed explanation for all methods will be given as follows: B-mode: 3 ¼ hypoechoic parenchyma, 2 ¼ inconclusive, 1 ¼ isoechoic parenchyma; power Doppler (vascularity): 3 ¼ hypervascular (high), 2 ¼ indifferent with atypical pattern (moderate), 1 ¼ hypovascular (low); strain elastography in combination with TDI: 3 ¼ dark blue colored (hard) and a fully extinguished area in the TDI, 2 ¼ blue or green and a partially extinguished area in the TDI, 1 ¼ green or red (soft) without an extinguished area in the TDI; CEUS: 3 ¼ early and strong enhancement and early washout, 2 ¼ early enhancement but no washout, 1 ¼ isoechogenic enhancement. As a proof of concept, the US modalities were scored fairly subjectively. After mpUS and clear reidentification of anatomic landmarks (seminal vesicles and Denonvilliers fascia), an MRI-US fusion biopsy was performed (Fig. 1I,J), followed by a conventional TRUS-guided 10-core biopsy without MRI-US fusion.
111
Table 2. Ultrasonography modalityebased performances Ultrasonography Modalities B-mode Power Doppler Strain elastography/tissue Doppler imaging CEUS T2w MRI appearance during initial US fusion
Sensitivity
Specificity
PPV
NPV
0.73 0.63 0.8 0.85 0.92
0.67 0.62 0.63 0.8 0.49
0.62 0.55 0.61 0.75 0.57
0.78 0.7 0.82 0.88 0.89
Mean Score Standard Deviation 1.63 1.57 1.68 1.62 1.85
0.73 0.66 0.7 0.74 0.7
CEUS, contrast-enhanced ultrasonography; NPV, negative predictive value; positive predictive value; T2w, T2 weighted. Other abbreviations as in Table 1. Calculated sensitivities, specificities, PPV, and NPV for PCa prediction of single ultrasonography modalities applied on PI-RADS predefined lesions.
comprised 45.6% of the total cohort. A total of 316 targeted lesions were biopsied. The mean number of targeted biopsies per patient was 1.86 (0.55). The mean total number of biopsies including the subsequent 10 cores (total n ¼ 1690) was 11.81 (0.82) per patient. The lesion-based analysis showed a cancer detection rate for the target biopsy of 21.2% (67 of 316) and 11.8% (200 of 1690) for the random biopsy. The overall cancer detection rate was 42% (71 of 169 patients). From 71 PCa patients, 48 (67.6%) underwent radical prostatectomy, where 5 patients (7.05%) underwent radiation therapy, 3 patients (4.23%) received hormonal therapy, and 15 (21.1%) were included in an active surveillance program. In accordance with the Standards of Reporting for MRItargeted Biopsy Studies working group recommendations,15 we divided our cohort into a clinically insignificant group (Gleason score
