First round of targeted biopsies using magnetic resonance imaging/ultrasonography fusion compared with conventional transrectal ultrasonography-guided biopsies for the diagnosis of localised prostate cancer Pierre Mozer*¶, Morgan Rouprêt*,**, Chloé Le Cossec†, Benjamin Granger†, Eva Comperat‡, Arachk de Gorski*, Olivier Cussenot*,** and Raphaële Renard-Penna§ Academic Departments of *Urology, †Biostatistics, ‡Pathology and §Radiology, AP-HP, Hopital Pitié-Salpétrière, ¶UPMC University of Paris 06, Institut des Systèmes Intelligents et de Robotique and **University Pierre and Marie Curie, Univeristy of Paris 06, GRC5, ONCOTYPE-Uro, Institut Universitaire de Cancérologie, Paris, France
Objectives
Results
To assess the accuracy of magnetic resonance imaging (MRI)/transrectal ultrasonography (TRUS) fusion to guide first-round biopsies in the diagnosis of localised prostate cancer (PCa) in men with a prostate-specific antigen (PSA) ≤10 ng/mL.
A total of 152 men, whose median PSA level was 6 ng/mL, were included in the study. The proportion of positive cores was significantly higher with the targeted-core protocol than with the extended 12-core protocol (P < 0.001). The proportion of men with clinically significant PCa was higher with the targeted-core protocol than with the extended 12-core protocol (P = 0.03). The proportion of patients having at least one positive biopsy (targeted-core protocol) was significantly different among the Likert score categories (P < 0.001).
Patients and Methods A prospective study was conducted on men who met the following criteria: first-round biopsy, multiparametric MRI (mpMRI) showing a lesion with a Likert score ≥2 and a PSA 10 years of experience in prostate MRI (R.R.P.) [21,22]. The analyses were performed on the respective workstations (Extended WorkSpace; Philips Healthcare, and Syngovia; Siemens Healthcare). The tumour data were reported in a collection form and included the largest size (mm), Likert score and localisation according to the consensus criteria for the use of MRI in the diagnosis and staging of PCa, based on a scheme using 27 regions of interest [23]. The reader provided a score for each area using scores of 1–5 on the Likert scale. The largest lesion on the mpMRI was defined as the ‘index target lesion’. This lesion was graded on a Likert scale according to the European Society of Urogenital Radiology prostate MRI guidelines [24], as follows: score 1 = clinically significant PCa is highly unlikely to be present; score 2 = clinically significant PCa is unlikely to be present; score 3 = clinically significant PCa is equivocal; score 4 = clinically significant PCa is likely to be present; and score 5 = clinically significant PCa is highly likely to be present. The reader was instructed to decide among these scores for each region based on his overall impression because no further criteria for these Likert categories are provided in the European Society of Urogenital Radiology panel’s description [20,24,25]. MRI/TRUS-Fusion Biopsy Procedure Both the MRI targeted and standard extended 12-core biopsies were performed at our institution during the same session by the same person. The practitioner was blinded to the targets during the extended 12-core biopsies. We used the UroStation™ (Koelis, Grenoble, France) and a V10 ultrasound system (Medison, Seoul, Korea) with an end-fire three-dimensional transrectal transducer. The UroStation implements elastic registration to fuse the MRI and three-dimensional TRUS images and allows guiding and the recording of core localisations on the three-dimensional TRUS and MRI images [26]. MRI images were imported from a CD-ROM or our picture archiving and communication system (PACS) into the UroStation. The shape of the prostate was delineated on T2-weighted magnetic resonance images. The patients underwent the extended 12-core protocol (12 systematic cores independent of mpMRI results) followed by sampling of two or three cores in the index target lesion by the same operator. During the extended 12-core protocol, the clinician was not aware of the exact mpMRI target localisation; the target was visible on the UroStation screen only for targeted biopsies. Conventional spring-loaded guns with 18-G needles were used. The workflow is shown in Fig. 1. Pathological Findings All of the biopsies were examined by one senior pathologist (E.C.). Tumour differentiation was determined using the © 2014 The Authors BJU International © 2014 BJU International
51
Mozer et al.
Fig. 1 Full workflow of the MRI/TRUS-fusion biopsy procedure. A, MRI images. B, Definition of a target (red arrow: target localisation; yellow arrow: target selected). C, Targeted biopsies in the selected target (MRI and TRUS image segmentation).
MRI
MRI
MRI
A
MRI Target
B
TRUS
3D
C
Table 1 Patient characteristics. All patients, N = 152
Age, years Prostate volume, mL PSA, ng/mL PSA density, ng/mL/mL Time from MRI to biopsy, days
Mean
SD
Median
IQR
Range
63 44 6 0.18 46
5.7 21.8 1.7 0.09 49
63.7 38.5 6 0.16 30
59.3–67.5 30–55 5–7.9 0.11–0.23 16–51
50–76 15–135 4–10 0.05–0.50 1–282
IQR, interquartile range.
Gleason score: high (2–6), moderate (7) or poor (8–10). The cores were individually inked with different colours to mark the sites from which they were collected. The cassettes were soaked in glasses of Bouin solution for 1 s to fix the colours and then preserved in pots with 10% formalin. In the present study, clinically significant PCa [27] was defined as follows: at least one core with a Gleason score of 3 + 4 or 6 with a maximum cancer core length ≥4 mm. Statistical Analyses The results are presented according to the START recommendations [20]. The proportions of cores positive for clinically significant disease were compared using the chi-squared test. The proportions of men with clinically significant disease were compared using a McNemar chi-squared test for paired percentages. Median comparisons were made using Wilcoxon’s test. The agreement between the targeted-core and the extended 12-core protocols was measured using Cohen’s κ coefficient. Cohen’s κ CIs were calculated by bootstrap. The relationship between the Likert score and the Gleason score obtained with targeted biopsies was explored using linear regression. The relationship between the Likert score and having at least one positive biopsy using the targeted-core protocol was explored using Fisher’s exact
52
© 2014 The Authors BJU International © 2014 BJU International
test. Thresholds for the prostate volume and the size of the lesion on the mpMRI were defined using a recursive partitioning method. The prediction matrix was developed using a model equation estimating the probability for one patient to have clinically significant PCa for each combination of identified predictors. The 95% CIs of individual probabilities were calculated with 250 bootstrap replications after removal of the outliers. All tests were two-sided and performed with a 5% α-risk. The statistical analysis was performed using R software (www.r-project.org).
Results A total of 152 men were included in the present study. The general characteristics of the patients are shown in Table 1. The median (interquartile range [IQR]) patient age was 63.7 (59.3–67.5) years, PSA level was 6 (5–7.9) ng/mL, and prostate volume was 38.5 (30–55) mL. The median (IQR) PSA density was 0.16 (0.11–0.23) ng/mL/mL. The median time between mpMRI and biopsy was 30 (15–51) days. The characteristics of the two protocols, recommended by the START guidelines, are shown in Table 2. The median number of cores taken per prostate was 12 for the extended 12-core protocol vs 2 for the targeted-core protocol. The number of men diagnosed with clinically significant PCa was 56 with the
MRI/TRUS-fusion targeted vs standard TRUS-guided biopsy
Table 2 Characteristics of the two protocols. Biopsy protocol for all patients, N = 152 Extended 12-core protocol
Targeted-core protocol
12
2
66 52 17 6 9 2
70 49 14 8 9 2
66 30 56
70 16 66
Median number of cores taken per prostate Gleason score, n No PCa 3+3 3+4 4+3 4+4 4+5 Disease significance, n No PCa Clinically insignificant Clinically significant PCa, prostate cancer.
Table 3 Cross-tabulation of the number of men with clinically significant and clinically insignificant cancer detected (N = 152). Extended 12-core protocol, n (%)
Targeted-core protocol, n (%) No PCa Clinically insignificant PCa Clinically significant PCa
No PCa
Clinically insignificant PCa
58 (38.2) 4 (2.6) 4 (2.6)
11 (7.2) 9 (5.9) 10 (6.6)
Clinically significant PCa
1 (0.7) 3 (2) 52 (34.2)
PCa, prostate cancer.
Table 4 Cross-tabulation of Gleason scores obtained with the targeted-core protocol and the extended 12-core protocol (N = 152). Extended 12-core protocol, n (%)
Targeted-core protocol, n (%) No PCa Gleason 6 Gleason 7 Gleason 8 Gleason 9
No PCa
Gleason 6
Gleason 7
Gleason 8
Gleason 9
58 (38.2) 5 (3.3) 3 (2) 0 0
10 (6.6) 39 (25.7) 3 (2) 0 0
2 (1.3) 5 (3.3) 15 (9.9) 1 (0.7) 0
0 0 1 (0.7) 8 (5.3) 0
0 0 0 0 2 (1.3)
PCa, prostate cancer.
extended 12-core protocol vs 66 with the targeted-core protocol. Cross-tabulation of the number of men with clinically significant, clinically insignificant and no PCa detected with the targeted-core protocol against cancers detected with the extended 12-core protocol is shown in Table 3. Cohen’s κ coefficient for the agreement between the two protocols for the diagnosis of clinically significant, clinically insignificant or no PCa detected was 65% (95% CI 54–74%). Cross-tabulation of the number of men in each Gleason score obtained for the targeted biopsies against the
scores obtained for the standard biopsies is shown in Table 4. Cohen’s κ coefficient for the agreement between the two protocols to determine the Gleason score was 70% (95% CI 59–79%). Statistical comparisons of the two methods are shown in Table 5. The proportion of cores positive for clinically significant PCa was significantly higher with the targeted-core protocol than with the extended 12-core protocol (P < 0.001). The proportion of men with clinically significant PCa was higher with the targeted-core protocol than with the extended 12-core protocol (P = 0.03). The © 2014 The Authors BJU International © 2014 BJU International
53
Mozer et al.
Table 5 Comparisons of the two biopsy protocols (N = 152). Extended 12-core protocol
Targeted-core protocol
P
7.5 37 12 4 (2–8)
31 43 2 8 (5–10)
Objectives
Results
To assess the accuracy of magnetic resonance imaging (MRI)/transrectal ultrasonography (TRUS) fusion to guide first-round biopsies in the diagnosis of localised prostate cancer (PCa) in men with a prostate-specific antigen (PSA) ≤10 ng/mL.
A total of 152 men, whose median PSA level was 6 ng/mL, were included in the study. The proportion of positive cores was significantly higher with the targeted-core protocol than with the extended 12-core protocol (P < 0.001). The proportion of men with clinically significant PCa was higher with the targeted-core protocol than with the extended 12-core protocol (P = 0.03). The proportion of patients having at least one positive biopsy (targeted-core protocol) was significantly different among the Likert score categories (P < 0.001).
Patients and Methods A prospective study was conducted on men who met the following criteria: first-round biopsy, multiparametric MRI (mpMRI) showing a lesion with a Likert score ≥2 and a PSA 10 years of experience in prostate MRI (R.R.P.) [21,22]. The analyses were performed on the respective workstations (Extended WorkSpace; Philips Healthcare, and Syngovia; Siemens Healthcare). The tumour data were reported in a collection form and included the largest size (mm), Likert score and localisation according to the consensus criteria for the use of MRI in the diagnosis and staging of PCa, based on a scheme using 27 regions of interest [23]. The reader provided a score for each area using scores of 1–5 on the Likert scale. The largest lesion on the mpMRI was defined as the ‘index target lesion’. This lesion was graded on a Likert scale according to the European Society of Urogenital Radiology prostate MRI guidelines [24], as follows: score 1 = clinically significant PCa is highly unlikely to be present; score 2 = clinically significant PCa is unlikely to be present; score 3 = clinically significant PCa is equivocal; score 4 = clinically significant PCa is likely to be present; and score 5 = clinically significant PCa is highly likely to be present. The reader was instructed to decide among these scores for each region based on his overall impression because no further criteria for these Likert categories are provided in the European Society of Urogenital Radiology panel’s description [20,24,25]. MRI/TRUS-Fusion Biopsy Procedure Both the MRI targeted and standard extended 12-core biopsies were performed at our institution during the same session by the same person. The practitioner was blinded to the targets during the extended 12-core biopsies. We used the UroStation™ (Koelis, Grenoble, France) and a V10 ultrasound system (Medison, Seoul, Korea) with an end-fire three-dimensional transrectal transducer. The UroStation implements elastic registration to fuse the MRI and three-dimensional TRUS images and allows guiding and the recording of core localisations on the three-dimensional TRUS and MRI images [26]. MRI images were imported from a CD-ROM or our picture archiving and communication system (PACS) into the UroStation. The shape of the prostate was delineated on T2-weighted magnetic resonance images. The patients underwent the extended 12-core protocol (12 systematic cores independent of mpMRI results) followed by sampling of two or three cores in the index target lesion by the same operator. During the extended 12-core protocol, the clinician was not aware of the exact mpMRI target localisation; the target was visible on the UroStation screen only for targeted biopsies. Conventional spring-loaded guns with 18-G needles were used. The workflow is shown in Fig. 1. Pathological Findings All of the biopsies were examined by one senior pathologist (E.C.). Tumour differentiation was determined using the © 2014 The Authors BJU International © 2014 BJU International
51
Mozer et al.
Fig. 1 Full workflow of the MRI/TRUS-fusion biopsy procedure. A, MRI images. B, Definition of a target (red arrow: target localisation; yellow arrow: target selected). C, Targeted biopsies in the selected target (MRI and TRUS image segmentation).
MRI
MRI
MRI
A
MRI Target
B
TRUS
3D
C
Table 1 Patient characteristics. All patients, N = 152
Age, years Prostate volume, mL PSA, ng/mL PSA density, ng/mL/mL Time from MRI to biopsy, days
Mean
SD
Median
IQR
Range
63 44 6 0.18 46
5.7 21.8 1.7 0.09 49
63.7 38.5 6 0.16 30
59.3–67.5 30–55 5–7.9 0.11–0.23 16–51
50–76 15–135 4–10 0.05–0.50 1–282
IQR, interquartile range.
Gleason score: high (2–6), moderate (7) or poor (8–10). The cores were individually inked with different colours to mark the sites from which they were collected. The cassettes were soaked in glasses of Bouin solution for 1 s to fix the colours and then preserved in pots with 10% formalin. In the present study, clinically significant PCa [27] was defined as follows: at least one core with a Gleason score of 3 + 4 or 6 with a maximum cancer core length ≥4 mm. Statistical Analyses The results are presented according to the START recommendations [20]. The proportions of cores positive for clinically significant disease were compared using the chi-squared test. The proportions of men with clinically significant disease were compared using a McNemar chi-squared test for paired percentages. Median comparisons were made using Wilcoxon’s test. The agreement between the targeted-core and the extended 12-core protocols was measured using Cohen’s κ coefficient. Cohen’s κ CIs were calculated by bootstrap. The relationship between the Likert score and the Gleason score obtained with targeted biopsies was explored using linear regression. The relationship between the Likert score and having at least one positive biopsy using the targeted-core protocol was explored using Fisher’s exact
52
© 2014 The Authors BJU International © 2014 BJU International
test. Thresholds for the prostate volume and the size of the lesion on the mpMRI were defined using a recursive partitioning method. The prediction matrix was developed using a model equation estimating the probability for one patient to have clinically significant PCa for each combination of identified predictors. The 95% CIs of individual probabilities were calculated with 250 bootstrap replications after removal of the outliers. All tests were two-sided and performed with a 5% α-risk. The statistical analysis was performed using R software (www.r-project.org).
Results A total of 152 men were included in the present study. The general characteristics of the patients are shown in Table 1. The median (interquartile range [IQR]) patient age was 63.7 (59.3–67.5) years, PSA level was 6 (5–7.9) ng/mL, and prostate volume was 38.5 (30–55) mL. The median (IQR) PSA density was 0.16 (0.11–0.23) ng/mL/mL. The median time between mpMRI and biopsy was 30 (15–51) days. The characteristics of the two protocols, recommended by the START guidelines, are shown in Table 2. The median number of cores taken per prostate was 12 for the extended 12-core protocol vs 2 for the targeted-core protocol. The number of men diagnosed with clinically significant PCa was 56 with the
MRI/TRUS-fusion targeted vs standard TRUS-guided biopsy
Table 2 Characteristics of the two protocols. Biopsy protocol for all patients, N = 152 Extended 12-core protocol
Targeted-core protocol
12
2
66 52 17 6 9 2
70 49 14 8 9 2
66 30 56
70 16 66
Median number of cores taken per prostate Gleason score, n No PCa 3+3 3+4 4+3 4+4 4+5 Disease significance, n No PCa Clinically insignificant Clinically significant PCa, prostate cancer.
Table 3 Cross-tabulation of the number of men with clinically significant and clinically insignificant cancer detected (N = 152). Extended 12-core protocol, n (%)
Targeted-core protocol, n (%) No PCa Clinically insignificant PCa Clinically significant PCa
No PCa
Clinically insignificant PCa
58 (38.2) 4 (2.6) 4 (2.6)
11 (7.2) 9 (5.9) 10 (6.6)
Clinically significant PCa
1 (0.7) 3 (2) 52 (34.2)
PCa, prostate cancer.
Table 4 Cross-tabulation of Gleason scores obtained with the targeted-core protocol and the extended 12-core protocol (N = 152). Extended 12-core protocol, n (%)
Targeted-core protocol, n (%) No PCa Gleason 6 Gleason 7 Gleason 8 Gleason 9
No PCa
Gleason 6
Gleason 7
Gleason 8
Gleason 9
58 (38.2) 5 (3.3) 3 (2) 0 0
10 (6.6) 39 (25.7) 3 (2) 0 0
2 (1.3) 5 (3.3) 15 (9.9) 1 (0.7) 0
0 0 1 (0.7) 8 (5.3) 0
0 0 0 0 2 (1.3)
PCa, prostate cancer.
extended 12-core protocol vs 66 with the targeted-core protocol. Cross-tabulation of the number of men with clinically significant, clinically insignificant and no PCa detected with the targeted-core protocol against cancers detected with the extended 12-core protocol is shown in Table 3. Cohen’s κ coefficient for the agreement between the two protocols for the diagnosis of clinically significant, clinically insignificant or no PCa detected was 65% (95% CI 54–74%). Cross-tabulation of the number of men in each Gleason score obtained for the targeted biopsies against the
scores obtained for the standard biopsies is shown in Table 4. Cohen’s κ coefficient for the agreement between the two protocols to determine the Gleason score was 70% (95% CI 59–79%). Statistical comparisons of the two methods are shown in Table 5. The proportion of cores positive for clinically significant PCa was significantly higher with the targeted-core protocol than with the extended 12-core protocol (P < 0.001). The proportion of men with clinically significant PCa was higher with the targeted-core protocol than with the extended 12-core protocol (P = 0.03). The © 2014 The Authors BJU International © 2014 BJU International
53
Mozer et al.
Table 5 Comparisons of the two biopsy protocols (N = 152). Extended 12-core protocol
Targeted-core protocol
P
7.5 37 12 4 (2–8)
31 43 2 8 (5–10)
