Automatic analysis of nuclear features reveals a non-tumoral predictor of tumor grade in bladder cancer

Abstract

Background & objectives: Tumor grade determines prognosis in urothelial carcinoma. The classification of low and high grade is based on nuclear morphological features that include nuclear size, hyperchromasia and pleomorphism. These features are subjectively assessed by the pathologists and are not numerically measured, which leads to high rates of interobserver variability. The purpose of this study is to assess the value of a computer-based image analysis tool for identifying predictors of tumor grade in bladder cancer.

Methods: Four hundred images of urothelial tumors were graded by five pathologists and two expert genitourinary pathologists using a scale of 1 (lowest grade) to 5 (highest grade). A computer algorithm was used to automatically segment the nuclei and to provide morphometric parameters for each nucleus, which were used to establish the grading algorithm. Grading algorithm was compared to pathologists' agreement.

Results: Comparison of the grading scores of the five pathologists with the expert genitourinary pathologists score showed agreement rates between 88.5% and 97.5%.The agreement rate between the two expert genitourinary pathologists was 99.5%. The quantified algorithm based conventional parameters that determine the grade (nuclear size, pleomorphism and hyperchromasia) showed > 85% agreement with the expert genitourinary pathologists. Surprisingly, the parameter that was most associated with tumor grade was the 10th percentile of the nuclear area, and high grade was associated with lower 10th percentile nuclei, caused by the presence of more inflammatory cells in the high-grade tumors.

Conclusion: Quantitative nuclear features could be applied to determine urothelial carcinoma grade and explore new biologically explainable parameters with better correlation to grade than those currently used.

Fig. 1

Examples of tumors with score 1 (a), the lowest grading score, and score 5 (b), the highest grading score

Fig. 2

a-c: The distribution of the grading scores of the five pathologists in comparison with the first (a), second (b) expert genitourinary pathologists, and the mean of the two experts (c). The horizontal scale represents the delta between the score of the pathologists and the expert pathologist. The agreement rates ranges between 88.5–97.5%. The graphs of the 5 pathologists “shifted to the right”, a finding that indicates that the pathologists tended to overgrade tumors in comparison to the expert genitourinary pathologists. Comparison between the two experts (d) showed very high agreement rate of 99.5%

Fig. 3

The computational analysis based nuclear features compared to the grading score of the pathologists. Tumors with grade score 1 (lowest grade) had higher mean nuclear area compared to tumors with grade score 3 and 5 (a). There was also a correlation between higher grade score and higher levels of standard deviation of nuclear area, a parameter that represents the nuclear pleomorphism (b). The parameter of the intensity of the color in the nuclei is seen (c), where lower levels on the scale indicate darker color. There was a correlation between higher grade score and darker nuclear color, a parameter that represents hyperchromasia

Fig. 4

The 90th percentile of nuclear area. There was a correlation between higher tumor grade score and higher 90th percentile of nuclear area

Fig. 5

The parameter of the 10th percentile of nuclear area compared to the grading scores of the pathologists. There was a correlation between higher grade score and lower 10th percentile of nuclear area (a), a finding that was explained by the presence of intense inflammatory cells infiltrating between tumor cells (b)

Fig. 6

Comparison of the automated nuclear parameters of nuclear area (a and c) and color intensity (b and d) with the grading scores of the two expert genitourinary pathologists showed agreement rates of > 85%

Fig. 7

Example of a tumor with high grade nuclear features with no inflammatory cells. Such images showed disagreements between the algorithm and the expert genitourinary pathologists in the parameter of the 10th percentile of the nuclear area

Fig. 8

Two examples of cases showing the grading scores of the five pathologists in comparison to the grade score of the expert genitourinary pathologists. There was a tendency among the five pathologists to overgrade tumors compared to the expert pathologists