| ObjectiveTo Quantitative analysis of the image analysis parameters of normal brain tissue and different grades of brain glioma cells and the parameters of the proliferation index Ki-67 and CyclinD1 expression,reveal the differences in morphology and density of different grades of glioma cellsand to screen valuable parameters for grading diagnosis of gliomas.Multiple glioma classification diagnosis models were established by linear discriminant function,and compared the diagnostic model of the back to determine the accuracy.Explored the significance of these parameters in the differential diagnosis of gliomas and provided an objective method for the histological grade of gliomas.Materials and Methods1.Experimental materials and groups:A total of 80 pathological specimens of brain gliomas were collected from Tongjiang Hospital from 2010 to 2015.All cases were untreated patients and no adjuvant or chemotherapy was performed before surgery.According to 2016 WHO histological grading criteria of gliomas,according to the case,the subjects were divided into control group and lesion group,control group of normal brain tissue,collected in 2010-2015 for brain trauma surgery for 10 cases of normal brain tissues.And the lesion group consisted of grade ?-? glioma group,including grade ? glioma in 15 cases,grade ? in 20 cases,grade ? in 25 cases,grade ? in 20 cases.2.Experimental method(1).Data acquisition and image analysis were performed by optical microscope,micro camera and image analysis software(Image-Pro Plus 6.0).At first,the test software was calibrated using the objective micrometer,and then HE sections were placed under the optical microscope objective lens with 40 times magnification to capture the microscopical fields randomly.1-4 microscopical fields for each sample were taken by the microscopic imaging system and input into the computer.Plus Image-Pro 6.0 image analysis software was used to measure the morphological parameters of the nuclei of different grades of gliomas.The morphological parameters included area of nucleus,majoraxis of nucleus,minoraxis of nucleus,and perimeter of nucleus.The cell number in these microscopical fields under the objective lens with 40 times magnification was counted,and then cell number density and nuclear area density were calculated,respectively.(2).Immunohistochemical method was used to detect the expression of Ki67 and CyclinD1 in WHO grade ?,grade ?,grade ?,grade ? gliomas and normal brain tissues.All specimens were routinely dehydrated and stained.After slicing and dewaxing,Ki67 and CyclinD1 were detected by immunohistochemical method.The brown yellow granules were positive in the nucleus of glioma cells,and the results were measured and recorded by the semi quantitative and 12 point method.Immunohistochemical staining of all specimens were set up positive control and negative control.The results of immunohistochemical staining were determined by two experienced clinicopathologists without knowing the results of the diagnosis.(3).SPSS16.0 statistical software was used for statistical analysis.The results of continuous variables were expressed as mean ± standard deviation(X),standard deviation(SD),maximum value(Min),and all measured values(X± SD).The test results of different groups were tested for the rank sum test of multiple independent samples.The comparative chi-square test of quantitative data of categorical data showed significant difference with P<0.05.Stepwise discriminant analysis of different model groups was conducted and Fisher linear discriminate function was established respectively.The discriminant function classification map was drawn and the internal-validation accuracy of each model group were calculated.Results1.Difference of nuclear morphological parameters in different grades of gliomacellsThe morphological parameters of the nucleus of grade ?-?gliomas showed a statistically significant difference in the area,major axis,minor axis,and perimeter.The average value of each morphological parameter of grade ? glioma was the highest among all grades,while the standard deviation was the lowest among all grades.The average morphological parameters(except for the major axis of nucleus)of Grade ? glioma were the smallest among all gliomas.The standard deviation of each morphological parameter of grade ?glioma was the largest in all gliomas.Low-grade gliomas(WHO grade ?-?)were significantly lower than those of high-grade gliomas(WHO grade ?-?)in area of nucleus and perimeter of nucleus.However,there was no significant difference in major axis of nucleus between grade ?and ? tumor(P>0.05),which regardless of area or perimeter.In addition to WHO grade ?and? tumor cells nucleus major axis of nucleus no significant difference,the other two groups were all statistically significant(P values were less than 0.05),suggested that low-grade gliomas usually had smaller nuclei.It might be an important image analysis feature to identify different histological grade of gliomas.Another feature was variation of nuclear morphology of gliomas.The variability of nuclear parameters in low-grade glioma was much smaller than that in high-grade gliomas,which suggested that there was obviously varied nuclear size of tumor cell in high-grade gliomas.It might be a factor caused by the polymorphism of tumor cells.2.Differences in density parameters of glioma cells in different gradesAlthough there was overall significant difference in cell number density of different grades of gliomas(P<0.001),there was no difference among higher-grade gliomas(WHO grade ?-?).It suggested that WHO grade ? tumor usually presented smallest cell number density,however,there was no significant difference in cellularity observed in higher-grade gliomas.Moreover,there were significant differences in the nuclear area densities of different grades of glioma cells(P<0.001),and there also were significant differences among the tumor with higher grade,suggested that the nuclear area density could be used as critical diagnostic clue to determine the accurate histological grading of gliomas.3.Difference of nuclear morphologic parameters and density parameters between normal brain tissue and gliomaThere was no significant difference in area of nucleus,minor axis of nucleus and perimeter of nucleus between normal brain tissue and WHO grade ? glioma in terms of nuclear parameters,except for the long axis of nucleus(P<0.01).And there was a significant difference in the area of nucleus,the major axis of nucleus,minor axis of nucleus and perimeter of nucleus between normal brain tissue and grade ?-?gliomas(P<0.01),suggesting that the nuclei of grade ?-?gliomas were significantly larger than those of normal brain tissues.In terms of cell density parameters,there was no significant difference in density parameters between normal brain tissue and grade I glioma,but there was significant statistical difference with grade ?-? glioma.It was suggested that the nucleus morphology and cell density of normal brain tissue are similar to that of grade ? glioma,but it was different from that of grade ?-? gliomas.4.Expression of Ki-67 in normal brain tissue and different grade gliomasThe expression of Ki-67 in gliomas is located in the nucleus and is yellowish to brown.No expression was found in 10 normal brain tissues,while the positive rate of Ki-67 in 80 cases of glioma was 67.50%(54/80),which was much higher than that of normal brain tissue(P<0.01).The expression level of Ki-67 in different grades of gliomas was not consistent.The expression level of Ki-67 was the highest in grade ?gliomas,and the lowest in grade ? gliomas.The Spearman rank correlation coefficient between Ki-67 expression score and ?-? glioma was 0.703(P<0.001),it showed that the level of Ki-67 index scores gradually increased with WHO grade ?-? glioma grade increased;In the group comparison,Ki-67 expression in low-grade gliomas(grade ?-? gliomas)and high-grade gliomas(grade ?-? gliomas)were significantly different(P<0.01);The expression of Ki-67 increased in grade ? and grade ? gliomas,but the difference was not statistically significant(P>0.05),similar to that in grade ?and grade ? gliomas.Among all the clinicopathological parameters,the expression of Ki-67 in different age groups was statistically significant(P<0.05),but there was no significant difference in gender,tumor size and location(P>0.05).5.Expression of CyclinDl in normal brain tissue and different grade gliomasThe expression of CyclinD1 in normal brain tissue and different levels of glioma was inconsistent.Only 1 case of CyclinD1 were positive in normal brain tissue(1/10),but the positive rate of expression in glioma was up to 83.75%(67/80),with significant statistical difference.The expression of CyclinD1 was the lowest in grade? glioma,and the highest in grade ? glioma.The Spearman rank correlation coefficient between the expression of CyclinD1 and grade glioma was 0.616(P<0.001),indicated that with the glioma level increased,the expression of CyclinDl gradually increased,and the staining intensity gradually increased.Moreover,in the expression rate of group comparison,?-? glioma differences were statistically significant(P<0.05).There was no significant difference in the expression of CyclinDl in different gender,age,tumor size and location(P>0.05).6.Correlation analysis of expression of Ki-67 and CyclinD1 in gliomaPearson correlation test showed that there was a significant correlation between Ki-67 and CyclinD1 expression in glioma,suggesting a significant positive correlation between the two proteins(R=0.816,?2=19.218,P<0.01).7.Multiple glioma grading diagnostic models based on image analysis parameters and proliferative index expression fractional Parameters(1).Based on morphological parameters and density parameters of normal brain tissues and different grades of gliomas,and the expression of Ki-67 and CyclinD1 were selectively added to perform Fisher's linear discriminant analysis.Four diagnostic models were obtained by establishing a linear discriminant function,which were the four-level glioma stereological diagnostic model,the four-level glioma image analysis + Ki-67 diagnostic model,the four-level glioma image analysis +CyclinD1 diagnosis Model and four-grade glioma image analysis + Ki-67,CyclinD1 diagnostic model,and the total internal-validation accuracy were 76.7%,75.6%,78.9%and 78.9%respectively.The total internal-validation accuracy rate of the diagnostic models were compared between groups,and the results showed no significant difference(P value was greater than 0.05).(2).The image analysis of low-grade gliomas(grade ?-? gliomas)and high-grade gliomas(grade ?-? gliomas)were integrated in the same time,and then add the normal brain tissue parameters of image analysis Fisher linear discriminant analysis,the diagnostic model of high and low grade gliomas was established by establishing a linear discriminant function,and the total internal-validation accuracy was 88.9%.Conclusions1.The morphological parameters such as area,major axis,minor axis,and perimeter in the glioma nuclei image analysis,and the number of cells per unit area and the nuclear area density,could serve as important index parameters for quantitative quantitative diagnosis of glioma histology,which were of great significance for the diagnosis and grading of glioma.2.The expression score of proliferation index factor Ki-67 and CyclinDl increased with the increase of glioma grade.The combination of the two indexes has a strong objectivity and practical value in the diagnosis of pathological grade and prognosis of glioma.3.Based on the image analysis of the nuclear area,nuclear major axis,nuclear minor axis,nuclear perimeter and unit area in the number of cells and nuclear area density,Ki-67 and CyclinD1 expression scores,multiple glioma grading diagnosis models could be established,and the total internal-validation accuracy of these models were similar(P>0.05),up to 78.9%..4.Based on the nucleus area,nuclear major axi,nuclear minor axis and nuclear perimeter of glioma cells,and the number of cells per unit area and the nuclear area density,and established a grading diagnostic model for high and low grade gliomas,the total internal-validation accuracy rate as high as 88.9%. |
PDF Full Text Request