Objective malignancy grading: a review emphasizing unbiased stereology applied to breast tumors

Low reproducibility reduces the clinical value of morphologic grading of malignant tumors, and the replacement of subjective classification by objective quantification has been suggested. Simple mitosis counting has been employed for objective malignancy grading most frequently and has proved its prognostic significance in, e.g., sarcomas and carcinomas of the breast and ovary. These and other measurements of morphometry are, however, obtained in two dimensions only, introducing bias due to ignorance of the fact that biologic structures are three-dimensional. Stereologic estimators are, to that end, well-suited, because they enable the assessment of spatial structure from sections. Studies addressing the impact of stereology in tumor pathology are the subject of the current review. Details of estimation are provided of stereologic variables of tumor size, numbers and densities of cancer cell nuclei and mitoses, mean size and size variability of cancer cell nuclei and variables of tissue architecture. Besides a description of their practical estimation the influence on variables of sampling method, tissue processing and observer variability is assessed, and estimator efficiency and measuring equipment is evaluated. Exemplifying the clinical importance of objective grading, results are summarized of prognostic studies of quantitative histopathology in women with breast cancer. It has been shown that many stereologic estimators are applicable to ordinary histologic sections processed under routine conditions. If a systematic random scheme of sampling is employed then the efficiency of estimation is usually high, and reproducible, accurate and representative results are ensured. For objective malignancy grading of breast cancer especially the volume-weighted mean nuclear size, vv (nuc), seems valuable, and the variable usually provides independent information to that of staging parameters. The prognostic value of vv (nuc) seems greatest in lymph node positive subsets, whereas the importance in lymph node negative patients should be further investigated. The clinical significance of some stereologic variables may be restricted due to relatively time consuming measurement procedures. However, the unbiased technique may provide precise measures of basic parameters like "tumor burden" and tumor growth pattern, and thereby be highly useful in experimental oncology. In conclusion, stereology is of great value for quantifying tumor elements. For objective malignancy grading especially assessment of the three-dimensional mean nuclear size seems useful. Prognostic significance of this variable has been demonstrated in, e.g., malignant melanoma and carcinomas of the breast, lung, bladder, prostate and uterine cervix. To determine the real clinical value of the measurements, further evaluation in a routine setting is necessary. In case such prospective studies confirm previous findings, the future replacement of subjective grading techniques by reproducible, objective variables seems feasible.