Histopathologic Techniques Gregorios Ebook Download
In this study, we used a restricted definition of lipids as the amount of lipid content detected in the tumor. That means the absence of lipid was used in the selection of patients to be included. We should have defined the type of the lipid in order to calculate the absolute amount. However, in daily practice, lipids are detected in quite a small amount of the tumor. Therefore, we concluded that it was reasonable to exclude patients with a non-lipid-containing tumor. Nevertheless, an earlier study reported that the amount of lipids in tumor demonstrated with various MR techniques significantly correlated with the tumor grade [ 3 ], which might explain the significant association we found between a non-lipid containing tumor and low rADC. In addition, strong correlation between tumor grade and proliferative activity determined by Ki-67 index was reported in an earlier study [ 1 ], in which the investigators calculated the Ki-67 index from biopsy or neoplastic cells identified using immunohistochemical staining of formalin-fixed, paraffin-embedded sections. Additionally, the Ki-67 value was associated with the prognosis of malignant gliomas [ 24 ], and the sensitivity and specificity for grading tumors were 100% and 87%, respectively [ 25 ].
From the foregoing, we can infer that the P value of the effects of neighborhood deprivation indicators on tumor grade, or neighborhood social and environmental factors on stage of disease at diagnosis, is not significant, but this is because the spatial random effects had a stronger association with grade/stage of disease. A similar concept applies to association with clinical and pathological predictors of tumor grade. With the spatial random effects, variance was reduced from 6.7% to 1.4% for grade, and from 10% to 1.3% for stage of disease. The spatial random effects were significantly related to stage of disease. The rate of disease at stage was elevated in blocks with higher concentrations of young, white, affluent women. The overall conclusion of this analysis is that there is considerable geographic variation in the grade and stage at diagnosis of prostate cancer, which was largely unexplained by a combination of individual-level and environmental factors. This study demonstrates the added value of using advanced statistical techniques to identify geographic variation in cancer outcomes and to identify areas of unexplained variation that may prove to have important etiologic implications.
The main limitation of our study was the use of only one ultrasound machine, which did not provide equally high accuracy, as in previous studies [ 2 ]. We found that the diagnostic performance of the first echography machine was limited (sensitivity 63.6%, specificity 84.5%) compared with that of the other machines (sensitivity 88.9%, specificity 95.9%). Omission of the first echography may have overestimated the accuracy of echography in our study. It seems that the first machine provided variable accuracy, as its diagnostic performance was higher when the final diagnosis was grade IV gliomas (sensitivity 100%) than grade III gliomas (sensitivity 50%; specificity 100%), and its predictive value was higher when excluding both low- and high-grade gliomas from the pathological analysis (sensitivity 91.7%, specificity 83.3%). Therefore, we decided to omit the first echography machine, which lacked consistency in its diagnostic performance and in the data gathered. However, several other studies that evaluated the performance of echography have considered only the first machine [ 40 ],[ 41 ],as the final diagnosis was confirmed with other diagnostic tests or other available data. We found that this machine provided adequate performance with a sensitivity of 91.7%, and we used this as the reference standard. A second limitation was our use of CE and CEA staining for immunohistochemical analysis. Previous studies have reported that CE3B and CEA are not prognostic factors in gliomas, in contrast to NSE, S-100, and GFAP [ 42 ]. However, as the specimens were collected in 2004, we selected markers of which the prognostic value has been previously described. Furthermore, CE3B, CEA, and GFAP staining appeared to be complementary, with each marker being suggestive of different histological features, confirming the validity of our results. Given the power of this study, it is unlikely that these factors could have influenced our findings. The known prognostic value of the Ki-67 antigen, although not significant in our analysis, is supported by the consistency of our results with previous studies. This factor could be an alternative in the assessment of grade of gliomas and it is not expected to be prognostic in most cases of GBMs (10% of GBM cases). It seems that tumor grade and Ki-67 could be complementary in assessing malignancy. The low accuracy and the lower predictive value for tumor grade of rCBV (31%) in comparison with other MRI techniques analyzed are in concordance with previous studies [ 43 ] and with the reduced sensitivity (11%) reported in other studies when including only grade IV gliomas [ 2 ],[ 44 ]. It seems that rCBV is highly reliable in terms of differentiation between low and high grade gliomas, but not between high and low grade gliomas. As MBP is diffusely expressed in all glial cells and has been reported as a marker of glial proliferation [ 45 ],[ 46 ], we believe that this positive expression may make some cases of low-grade gliomas appear more hypercellular than others, thus increasing the accuracy of diagnosis of grade III gliomas by MBP; however, our study did not demonstrate this effect. Furthermore, the high accuracy of rADC could be related to the fact that this parameter is differentially expressed according to the type of cell. It appears that increased amounts of lipids may increase the cellularity of glial cells, enhancing the contrast of tumors. However, our study found a reduction of cellularity and lipids when calculating this parameter. It seems that rADC was not sensitive enough for the evaluation of this parameter.