To assess image quality and accuracy of the prostate MRIs performed at our center to serve as an internal audit in the detection of clinically significant prostate cancer.
All male pelvic genitourinary MRIs performed at our center between 1st January 2021 and 31st December 2022 were retrospectively reviewed. Prostate MRIs were selected (Figure 1) and image quality was assessed according to PI-QUAL and the PI-RADS v2.1 category was reviewed. Patient records were revised to determine if a biopsy was performed and the histopathologic result was used as a gold standard for determining the diagnostic accuracy of imaging studies performed. A descriptive analysis of our findings is given.
Figure 1. Genitourinary MRIs performed between 1st January 2021 and 31st December 2022 - included and excluded cases.
Figure 2. Pie chart depicting the PI-QUAL distribution of prostate MRIs performed
Figure 3. Pie chart depicting the PI-RADS v2.1 distribution of prostate MRIs performed
A total of 198 male pelvic genitourinary MRIs were performed in the period between 1st January 2021 and 31st December 2022 (Figure 1), of which 187 evaluated the prostate (pre-treatment). Most of our studies (96%) were of sufficient diagnostic quality (PI-QUAL ≥3), with 81% being of optimal quality (PI-QUAL ≥4) (Figure 2). We do not have any PI-QUAL 5 studies because we are not able to comply on our equipment with the DWI requisite PI-RADS v2.1 of a high b value of minimum 1400. All our studies use a high b value of 1200. A negative MRI (PI-RADS 1 and 2) was found in 85 cases (45,5%), a positive MRI (PI-RADS 4 and 5) was diagnosed in 77 cases (41,2%) and an indeterminate MRI (PI-RADS 3) was identified in 25 cases (13,3%) (Figure 3). The distribution of image quality, according to PI-QUAL, for each type of MRI (negative, indeterminate, or positive) is given on Figure 4. A negative MRI (PI-RADS 1 and 2) was found in 85 cases (45,5%) (Figure 3) and biopsy was avoided in 74 patients (39,6% of the total number of MRIs performed). A positive MRI (PI-RADS 4 and 5) was diagnosed in 77 cases (41,2%) (Figure 3). Of these 67 had cancer (22 MRIs performed for staging after a positive biopsy and 45 were diagnosed at MRI). Three cases had a negative biopsy and a revision and re-biopsy was suggested - we present one of these cases which likely is associated with undersampling at biopsy due to the anterior location of the suspicious lesion (Figure 5). Nine cases did not perform biopsy (4 lost to follow up, 2 refused biopsy and 3 no reason given on clinical records). An indeterminate MRI (PI-RADS 3) was identified in 25 cases (13,3%) (Figure 3). Of these 11 underwent biopsy with 7 having chronic prostatitis while 4 had cancer. Regarding clinically significant cancers, considered as ISUP ≥2, these were identified in 57 cases (Figure 6), with the majority (93%) being identified on PI-RADS 4 and 5 cases. The 3 clinically significant cancers given a PI-RADS 2 on imaging were studies performed for staging after the histopathological diagnosis with no index lesion detected on imaging. Considering a positive MRI as PI-RADS 4 and 5 our accuracy was 88%, with a good sensitivity and specificity (89% and 81% respectively), while if considering a positive MRI as PI-RADS 3, 4 or 5 our accuracy was 84% with a rather low specificity (Figure 7). There are some limitations to this research, namely: (1) a single reader in all cases as there is only one dedicated genitourinary radiologists in the department; (2) not all PI-RADS 3, 4 and 5 lesions were biopsied for confirmation; and (3) the histopathological results are based on cognitive fusion biopsies at our center.
Figure 4. Pie charts depicting the distribution of PI-QUAL according to PI-RADS category
Figure 5. PI-RADS 5 lesion in the anterior fibromuscular stroma, on a PI-QUAL 4 study, that had a negative biopsy, likely due to undersampling. Images shown correspond to T2W sagittal, T2W axial, DWI and ADC map with blue arrows pointing the suspicious lesion.
Figure 6. Chart depicting clinically significant cancer distribution according to PI-RADS category.
Our findings are in accordance with those reported from expert centers with a low percentage of PI-RADS 3 cases and a good accuracy to diagnose prostate cancer.
Figure 7. Tables presenting the information of sensitivity, specificity, positive predictive value, negative predictive value and accuracy when considering a positive MRI test as PI-RADS (3, 4 or 5) or as PI-RADS (4 or 5)