Use of Histologic Grade as a Prognostic Variable.
Adjuvant! 5.0 provides the user with the opportunity to use histologic grade in the prognostic assessment of breast cancer patients. Histologic grade is widely but not universally used in the prognostic assessment of patients. If histologic grading information is not available, or if the histologic grading is felt to be unreliable, this parameter can be set to undefined and ignored.
In order to use a parameter as a prognostic variable it must have at minimum two characteristics: it must be reliably measurable and it must confer strong prognostic information that is statistically independent of other variables that are to be used.
Whether histologic grade can be reproducibly measured has been a point of some controversy (1). There are however major studies that show moderately good agreement in histologic grading between pathologists. Indeed the reproducibility may be as good as the measurement of tumor size and number of involved axillary nodes both of which we might expect considerable variability in the measurement of, and for which we have little or no information as to the reproducibility of measurement.
In an American study (2) a single slide from 10 cases of invasive breast cancer was submitted to 25 pathologists. These pathologists were given a set of guidelines for the widely used form of Bloom-Richardson grading system as modified by Elston (3). Tumors were rated as low, intermediate, or high grade depending on whether their combined B-R score was 3-5, 6 or 7, or 8 or 9 (see B-R Histologic Grading). In this study 216 of the 250 (86%) of the assigned combined grades were in agreement with the majority opinion. In 3 of the cases there was 100% agreement, and in 6 cases 87% agreement. The 1 case where there was strong discordance lay at the boundary between low and intermediate grades with about half the pathologists scoring the case as a 6 and most of the others scoring it as a 5. The authors concluded that pathologists using a simple set of guidelines could achieve good reproducibility.
In a Swedish study (4) slides from 93 invasive breast cancers were sent to 7 Swedish pathology departments. The publication does not specify whether all sites were given a set of guidelines of definitions of the difference parameters used in grading. In some ways the results were similar to the American results. In this study 534 of 651 (82%) of assigned grades were in agreement. In 31% percent of the cases there was agreement as the overall histologic grade at all centers. Overall this study showed, like the American study, moderate reproducibility.
One the studies that has shown the prognostic power of histologic grade is a previous study of this issue using SEER data from 1977-1986 that included 22,616 cases (5). This study is of particular interest because it was done in an era when very few node negative patients received systemic adjuvant therapy. The results of this analysis are shown below.
In a number of ways the results are similar but there are some important differences. The similarities are in the apparent numbers derived from the earlier analysis and that done for Adjuvant! version 3.0. There are some important differences however. For the earlier analysis it is not at all clear how they avoided the confounding effects of non-breast cancer mortality in patients older than 70. Unfortunately the earlier analysis was done using different cut off tumor sizes than that which are used today, and this makes the two sets of results not completely comparable. Nonetheless the results show the excellent prognosis of patients with small tumors and no axillary node involvement.
There are numerous other studies that show the importance of histologic grade. In an analysis of 613 Norwegian patients with T1N0 and T2N0 tumors with a median follow-up of greater that 15 years none of whom received multi-cycle chemotherapy, histologic grade was a powerful prognostic variable (6). In a multivariate analysis controlling for tumor size and histologic grade as predictors of death, T2 vs T1 conferred a 1.8 (1.3-2.4) fold risk and G1 vs G2/3 conferred a 2.4 (1.5-3.9) fold risk.
The Eastern Cooperative Oncology Group (7) addressed the question of the prognostic value of grade using samples from 368 node positive patients who had participated in their adjuvant protocols. They found that histologic grade remained a strong prognostic parameter (for both DFS and OS) even after adjusting for tumor size, number of nodes, and s-phase fraction. Because of the inclusion of S-phase fraction and mitotic index in some of the modeling it was difficult to discern the independent impact of histological grade on patient outcome for scenarios where only tumor size and number of nodes were known. It appeared to be approximately 1.5 and 1.7 for relapse low grade versus immediate and high grade respectively. There were no estimated hazard ratios given for the multivariate analyses of OS but the Kaplan Meier curves (really a univariate look at the data) suggest that the hazard rations may be approximately 2 for Grade 1 vs 2 and Grade 2 vs 3.
These studies indirectly show that grade must be reasonably reproducible, or it could not appear to be such a powerful independent predictor.
Clearly histologic grade is a powerful but somewhat subjective parameter. This has led to varying views on whether to use it. To some extent histological grading has been held to a higher standard than tumor size and estimates of number of positive nodes, both of which we have relatively little information about the precision of measurement, and for which there is probably more variability in measurement than generally appreciated. In the SEER data histologic grade is a very powerful prognostic parameter despite variation in its measurement.
(1) Page DL. Ellis IO. Elston CW. Histologic grading of breast cancer. Let's do it. American Journal of Clinical Pathology. 103(2): 123-124, 1995.
(2) Dalton LW, Page DL, and Dupont WD. Histologic grading of breast cancer. A reproducibility study. Cancer 73: 2765-2770, 1994.
(3) Elston CW. Grading of invasive carcinoma of the breast. In: Page DL, Anderson TJ, editors. Diagnostic histopathology of the breast. Churchill Livingston. 1987: 300-311.
(4) Boiesen P. Bendahl P-O, Anagnostaki L., et al. Histological grading in breast cancer. Acta Oncologica 39(1): 41-45, 2000.
(5) Henson DE. Ries L. Freedman LS. and Carriaga M. Relationship among outcome, stage of disease, and histologic grade for 22,616 cases of breast cancer.
(6) Reed W. Hannisdal E. Boehler PJ. et al. The prognostic value of p53 and c-erbB-2 immunostaining is overrated for patients with lymph node negative breast carcinoma. Cancer 88: 804-813, 2000.
(7) Simpson JF. Gray R. Dressler LG. et al. Prognostic value of histologic grade and proliferative activity in axillary node-positive breast cancer: Results from the Eastern Cooperative Oncology Group companion study EST 4189. Journal of Clinical Oncology 18 (10): 2059-2069, 2000.
Reviewed, but not revised 4/03