| Part Ⅰ:The clinical and pathological characteristics of false-negative prostate cancer lesion and region diagnosed by biparametric magnetic resonance imagingObjectiveTo investigate the sensitivity,and the clinical and pathological characteristics of falsenegative prostate cancer(PCa)lesion and region diagnosed by biparametric magnetic resonance imaging(bp-MRI).MethodsA total of 177 patients receiving radical prostatectomy(RP)during August 2017 and January 2022 were retrospectively studied.The mean age was 68.83±6.31 years,with median PSA level of 8.80(5.79-12.47)ng/ml,and the median prostate volume of 33.25(24.18-45.35)ml.All patients underwent bp-MRI.The whole mount section was prepared.Two doctors diagnosed all imaging according to the Prostate Imaging Reporting and Data System,version 2.1,without knowing the pathological results.One doctor completed all pathological diagnosis independently.The sensitivity was calculated on both the patient and lesion levels.The Logistic Regression analysis was conducted to select the associated factors.For lesions diagnosed correctly,pathological characteristics were recorded for lesions with invisible regions,including Gleason grade,location,and tumor density.Differeces with a P<0.05 was considered statistically significant.ResultsAmong all 177 patients,a total of 163 were correctly diagnosed with bp-MRI.The sensitivity was 92.1%for patient level.Atotal of 366 lesions was found on whole mount section,in contrast,227 lesions were detected on bp-MRI,with a sensitivity of 62.0%for lesion level.A total of 185(50.4%)International Society of Urological Pathology(ISUP)grade group 1lesions,113(30.8%)ISUP grade group 2 lesions,53(13.6%)ISUP grade group 3 lesions,15(4.1%)ISUP grade group 4 or higer lesions were detected by whole mount section.The maximum diameter,index lesion and ISUP grade group were significantly associated with PCa detection.Tumor located in the transitional zone,with lower density and lower ISUP grade group were more likely to be invisible region on bp-MRI.ConclusionBp-MRI has a high sensitivity and reliability for PCa diagnosis.The maximum diameter,index lesion and ISUP grade group were significantly associated with PCa detection.Tumor located in the transitional zone,with lower density and lower ISUP grade group were more likely to be invisible region on bp-MRI.Part Ⅱ:The accuracy of T2WI and ADC map on the localization of prostate cancer with different grade groupsObjectivesTo investigate the ccuracy of T2WI and ADC map on the localization of prostate cancer(PCa)with different grade groups,using the whole mount section as a reference standard.MethodsPatients diagnosed with PCa and undergoing radical prostatectomy during August 2017 and January 2022 were retrospectively studied.All patients underwent biparametric manetic resonance imaging(bp-MRI).The whole mount section was prepared,and was scanned as digital sections for further analysis.Two doctors diagnosed all imaging according to the Prostate Imaging Reporting and Data System,version 2.1,without knowing the pathological results,and segmented lesions with ITK-snap.One doctor completed all pathological diagnosis independently and segmented the lesion on digital sections.Then tumor volume was used.Tumor volume was calculated on T2WI,ADC map and digital sections.The linear regression model was used and the Pearson correlation analysis was conducted.Further,patients were divided into three groups named high-grade group,intermediate-grade group and low-grade group,according to the International Society of Urological Pathology(ISUP)grade groups.Tumor vulumes were calculated and compared to indentify the most accurate sequence in a specific group.Differeces with a P<0.05 was considered statistically significant.ResultsA total of 158 index lesions were found on bp-MRI.The tumor volume was 2.74±4.69 cm3,2.82±4.62 cm3,and 3.49 ± 5.22 cm3,respectively for T2WI,ADC map and digital sections.Both T2WI and ADC map significantly underestimate the tumor volume(P<0.001).The correlation coefficients were rT2=0.91 and rADC=0.93 for T2WI and ADC map.On subgroup analysis,the tumor volume was 1.29±0.95mL,2.62±3.90mL,and 9.31±11.20mL for T2WI in the low-,intermediate-,and high-grade gourps,respectively.The tumor volume was1.47±1.32mL,2.80±4.11mL,and 3.74±10.66mL for ADC map in the low-,intermediate-,and high-grade gourps,respectively.The tumor volume was 1.76± 1.27mL,3.69±4.99mL,and 8.05±11.12mL for digital sections in the low-,intermediate-,and high-grade gourps,respectively.The correlation coefficients were 0.52,0.91,and 0.98 for T2WI and digital sections(All P<0.001).In contrast,the correlation coefficients were 0.60,0.91,and 0.98 for ADC map and digital sections(All P<0.001).T2WI tends to overestimate lesion volume in high-grade group,and the ADC map tends to overestimate lesion volumes in low-grade gourp.ConclusionBoth T2WI and ADC map will significantly underestimate the tumor volume,which should cause attention in clinical practice.For PCa lesions with ISUP grade group 1-3,the tumor volume calculated by ADC map was loser to that calculated by digital sections;however,T2WI appears to be a more stable parameter.There is no optimal sequence for tumor volume estimation alone till now.A combined scheme using multi modalities should be used to make the plan for targeted biopsy and focal therapy.Part Ⅲ:Prediction of clinically significant prostate cancer based on a radiomic signaiture from biparametric magnetic resonance imagingObjectives:To build a model based on the radiomic signatures for the prediction of clinically significant prostate cancer(csPCa)and test the diagnostic efficiency.MethodsPatients receiving radical prostatectomy at our institution were retrospectively studied.All patients underwent biparametric magnetic resonance imaging(bp-MRI).Segmentation was conducted on T2WI and diffusion-weighted imaging(DWI),which was used for radiomic features extraction.Mann-Whitney U test and a LASSO regression analysis were conducted to select the most relavant features.A logistic regression model was used for the radiomic signature construction.Clinical factors including PSA,location,Prostate Imaging Reporting and Data System(PIRADS)score was analyzed via univariable and multivariable analysis.Futher,a clinical model and a combined model were builts.Receiver of Characteristics Curve(ROC)analysis and area under the curve(AUC)was conducted to test the diagnostic efficiency of the model.The calibration curve analysis was also conducted.Differences with a P<0.05 were considered statistically significant.ResultsA total of 112 index lesions were detected on bp-MRI.A total of 78 patients(70%)were randomly selected for the training cohort,and the rest(34 patients,30%)for the test cohort.A total of 17 features were selected for the radiomic signature construction.For clinical factors,PSA and location were the significantly associated factors by univariable and multivariable analysis.On ROC curve analysis,the AUC of the radiomic model was 0.937 and 0.661 for the training and test cohort,respectively.The AUC of the clinical model was 0.710 and 0.665 for the training and test cohort,respectively.The AUC of the combined model was 0.929 and 0.636 for the training and test cohort,respectively.ConclusionRadiomic features from bp-MRI could identify csPCa well,which might be helpful for the future preoperative diagnosis.Part Ⅳ:Diagnostic value of the use of 18F-prostate specific membrane antigen PET/CT combined with biparametric-MRI in identifying the grade group of prostate cancerObjectiveTo investigate the diagnostic value of the combination of 18F-prostate specific membrane antigen(PSMA)PET/CT and biparametric magnetic resonance imaging(bp-MRI)in identifying the grade group of prostate cancer,using parameters derived from the two imaging modalities.MethodPatients diagnosed of prostate cancer by histopathology and receiving 18F-PSMA PET/CT and bp-MRI during September 2018 to May 2019 in our hospital were retrospectively studied.The median age was 68(64-75),with the median PSA level of 14.74(7.75-24.19)ng/mL.All patients received mpMRI before biopsy.On biopsy,6(12.2%)patients had International Society of Urological Pathology grade group(ISUP GG)1 diseases,16(32.7%)had ISUP GG 2 diseases,12(24.5%)had ISUP GG 3 diseases,and 15(10.9%)had ISUP GG 4 or 5 diseases.Patients were then divided into high-grade group(ISUP 4-5)and lowgrade group(ISUP 1-3).The median age of patients in high-grade group and lowgrade group were 65(62-76)and 71(65-74),respectively.The PSA level in high-grade group and low-grade group were 15.11(6.63-42.86)ng/ml and 12.31(7.94-18.25)ng/ml,respectively.No significant differences were found in age and PSA level between the two groups(P=0.334,P=0.448).All patients underwent 18F-PSMA PET/CT within 4 weeks after biopsy.The maximum standardized uptake value(SUVmax)and the minimum apparent diffusion coefficient(ADCmin)were recorded,and the ratio of SUVmax/ADCmin were calculated.The correlation between the above parameters and ISUP grade group were analyzed.The diagnostic value of the parameters was evaluated by the receiver operating characteristic(ROC)curve.ResultsThe data of 49 patients were analyzed.The average ADCmin was(0.57±0.16)× 10-3 mm2/s,with the average SUVmax and SUVmax/ADCmin of 15.30±12.54 and(29.69±23.72)×103,respectively.Statistical differences were found in SUVmax(P=0.012)and SUVmax/ADCmin(P=0.002)between the high-and low-grade groups,while ADCmin(P=0.411)saw no statistical differences between the two groups.Significant positive correlations were found between SUVmax(r=0.501,P<0.001),SUVmax/ADCmin(r=0.527,P<0.001)and ISUP grade group,respectively.There was a negative correlation between ADCmin and ISUP grade group(r=-0.296,P=0.039).SUVmax/ADCmin was the best index to distinguish high-grade group from low-grade group prostate cancer with the area under the curve(AUC)of 0.749.In contrast,the AUC of SUVmax and ADCmin were 0.731 and 0.615,respectively.The diagnostic sensitivity and specificity of SUVmax/ADCmin were 73.3%and 85.3%,respectively,with a critical value of 37.23×103.ConclusionThe combination use of 18F-PSMAPET/CT and bp-MRI could improve the diagnostic efficiency for prostate cancer,compared to either modality alone.The ratio of SUVmax/ADCmin has a positive correlation with ISUP grade group,and is a promising index for diagnosing the high-grade prostate cancer from low-grade cancer. |
PDF Full Text Request