The Comparative Study Between MRSI And Pathological Specimens In Prostate Cancer

Objective:Making cross-check analysis and research between the MRSI of a person who has prostate cancer before the operation and the pathological result of isolated specimen of corresponding areas after the cancer is rooted out to evaluate a. the accuracy, sensitivity and specificity of the current MRSI diagnostic criteria to the diagnosis of prostate cancer, b. the relation among the CC/C value of MRSI, the type of the abnormal tissue and the tissue heterology, and c. the relativity among the CC/C value of MRSI, the Gleason score and the degree of abnormal differentiation. To test the reliability of MRSI diagnosing prostate cancer and to further discuss the predictive function of MRSI to the degree of abnormal differentiation and the Gleason score.Materials and Methods:Materials:5 cases who have prostate cancer and will go under the radical prostatectomy are needed. Each one accepts the MRI and the MRSI examination. Within thirty minutes after the operation, the fully cut prostate specimens accept the MRSI examination again. Then fix the specimens and slice them with the widest MRSI cross section. Taking measures of the size of the MRSI examination interesting area(ROI), the MRSI result and the pathological result of each inside interesting area(ROI) are compared and analyzed. Methods:a) MRI and 3D MRSI examination:The MRI and the MRSI scanning of all patients are done by GE Sigma 1.5T MRI, MRI scan include axial T1WI and T2WI, sagittal and coronal FS-T2WI.3D MRSI have 5-12 axial images in each case, and each slice contains 15-35 ROI. b) MRSI examination of the prostate specimens after operation:Within 30 minutes after the operation, the prostate specimens accept the MRSI examination again. After fixing the specimens, marked by the urethras, the seminal vesicles and the front part of the prostates, the prostate position before the operation scanning is simulated. Using the same coil, weight set in 3 kilogram and the parameter of the scan is the same as the scan before operation.c) Making the pathological section of prostate specimens:Put each prostate specimen in formaldehyde for 10 to 36 hours. The specimens are marked by the urethra and the seminal vesicles, barked on the front part of the prostate, compared with the widest cross section in MRI, select and slice center layer in specimens. Contrasting with the MRSI image, the sliced layer is divided into 15 to 35 equal pieces. Every piece is numbered according to the order of MRSI and prostate regional pathological sections are made.d) The diagnosis of the pathological sections of the prostate specimens:After the completion of the making and preparation of all pathological sections, two experienced doctors, who are ranked above associate chief physicians coming from Tianjin Urinary Institute, write the pathological report in double blind method, confirming the types of pathology tissues of each section and getting the Gleason score of each region.e) The diagnosis of MRSI:The data of MRSI of all patients are automatically processed at GE ADW4.2 work station by Functool software. The Cho,Cr,Cho/Cr, LL, Cit and CC/C values of all ROI in interesting layers are recorded. Using CC/C＞0.86 as the diagnostic standard of prostate cancer, two associate chief radiologist analyze and report the area which contain cancer in all ROI. f) The comparative study between the MRSI and the pathological result in the same area:According to the outcome of the comparative study between the MRSI and the pathological result in the same area, the research includes:1. counting the accuracy rate, the sensitivity and the specificity of MRSI diagnosing prostate cancer under the diagnostic standard of the CC/C 0.86; 2. the accuracy rate of MRSI diagnosis in different Gleason score groups and its relation with Gleason score; obtaining the cutoff value of these data to the diagnosis of prostate cancer through statistical calculating according to pathological and MRSI results; and analyzing the relativity between CC/C values and the types of pathology tissues and Gleason score.g) The statistical analysis methods:All statistical analyses are finished by SPSS11.5 with the examination levelα=0.05.Analysis of data was used to compare differences between groups of independent samples t test, paired t test, variance analysis, multiple comparisons using LSD method; the analysis of the correlation between indicators Spearman rank correlation analysis was completed using hypothesis testing; using the area under the ROC curve of this set of data to chose the best CC/C values as the diagnosis standard of prostate cancer and the estimating standard of the tumor differentiation.Results:a) The reliability study of MRSI to the diagnosis of prostate cancer when CC/C is above 0.88:90 effective ROI are obtained from 5 specimens. Through the pathological diagnosis,71 regions are recognized as cancer and 19 ones are recognized as non-cancer. According to the internationally recognized criterion of CC/C above 0.86, MRSI diagnosis figures out that 65 regions are cancer and 25 regions are non-cancer. Of these, the correct number of MRSI diagnosed prostate cancer is 59(59/65) and the correct number of MRSI diagnosed non-prostate cancer is 14(14/25). The result shows that with regard to MRSI diagnosing prostate cancer, the accuracy is 81.1%, the sensitivity is 84.3% and the specificity is 70.0%.b) Based on this pathological result, it is confirmed that the best cutoff value of CC/C diagnosis of prostate cancer is 0.88. The accuracy of this criterion is 80.0%; the sensitivity is 81.4%; and the specificity is 75.0%.c) The correlational study of the value of CC/C and pathological types of prostate cancer:In the 70 cancer regions, pathological types are acinosum carcinoma (39 cases), cribriform carcinoma (53 cases), infiltrating carcinoma (32 cases), solid carcinoma (12 cases) and mucinous carcinoma (2 cases). Their corresponding CC/C values are 1.563±0.998 for acinosum carcinoma, 1.808±1.081 for cribriform carcinoma,1.946±0.791 for infiltrating carcinoma, 2.509±1.261 for solid carcinoma and 2.515±0.924 for mucinous carcinoma. After conducting statistical analysis of above figures, it shows that the corresponding CC/C variance analysis result of the number of pathological type in the same area, represented by P, is above 0.05 and with no statistical meaning (F equals to 1.360 and P equals to 0.264). Among all the cancer regions, the number of the regions, which only have one pathological type, is 15. And the mean value of CC/C is 1.746±1.248. The number of the regions, which have two pathological types, is 42. The mean value of CC/C is 1.680±1.035. The number of the regions, which have three pathological types,is 13. The mean value of CC/C is 2.229±0.884. According to Spearman's relative analyses, the number of pathological types is in positive correlation with corresponding CC/C value. That is the heterology number of cancer tissues is in direct proportion to the CC/C value (r=0.419, P=0.000<0.05). d) The study of CC/C values and the pathology of prostate cancer:According to Spearman's relative analyses, the CC/C value increases proportionally to the Gleason score (r=0.746, P=0.000＜0.05) and has statistical meaning. According to CC/C value chose the Cutoff value (optimal threshold) of the differentiation of prostate cancer, all data were divided into 2 groups using the demarcation point as Gleason score 7. The Gleason score of well differentiated carcinoma group is less than or equaling to 7 score, using ROC curve analysis the cutoff value, the outcome show the lesion's Gleason score and CC/C value was no significant correlation (r=0.157, P=0.562> 0.05); poorly differentiated carcinoma group's Gleason score is all above 7, by analyzing the area under the ROC curve, choose 0.948 for the optimal threshold, its sensitivity was 81.4%, specificity is 75.0, in the other hand, in poorly differentiated carcinoma group, Gleason score and CC/ C values are related (r=0.605, P=0.000<0.05). The outcome of this study prompted greater the lesion is most like poorly differentiated carcinoma when the CC/C value is above 0.948.Conclusions:a) The set of data confirmed to, under the diagnostic standard as vCC/ C values> 0.86, the accuracy of diagnosis of prostate cancer by MRSI is 81.1%, sensitivity is 84.3%, specificity is 70.0%; the accuracy of diagnosis of MRSI increased with the rising of Gleason score which was significant correlation. b) By statistical analyzing the set of data, the best threshold value of CC/C diagnosing PCa is 0.88, the accuracy, sensitivity and specificity is 80.0%,81.4% and 75.0%, which result is similar to the outcome under CC/C> 0.86, and this standard of CC/C>0.88 may be more appropriate for Chinese people. c) In prostate cancer, CC/C value is positively correlated to both the number of pathological type and the Gleason score. d) CC/C values can be used to estimate the degree of differentiation of prostate cancer, when CC/C>0.948, it prompt for the possibility of poorly differentiated carcinoma and in poorly differentiated carcinoma group, the Gleason score shows positively correlation with the CC/C value.