Histologic Grade Remains a Prognostic Factor for Breast Cancer Regardless of the Number of Positive Lymph Nodes and Tumor Size: A Study of 161 708 Cases of Breast Cancer From the SEER Program

^ Context.-The appropriate staging of breast cancers includes an evaluation of tumor size and nodal status. Histologic grade in breast cancer, though important and assessed for all tumors, is not integrated within tumor staging.

Objective.-To determine whether the histologic grade remains a prognostic factor for breast cancer regardless of tumor size and the number of involved axillary lymph nodes.

Design.-By using a new clustering algorithm, the 10- year survival for every combination of T, N, and the histologic grade was determined for cases of breast cancer obtained from the Surveillance, Epidemiology, and End Results Program of the National Cancer Institute. There were 36 combinations of TN, defined according to the American Joint Committee on Cancer, and grade.

Results.-For each combination of T and N, a categor- ical increase in the histologic grade was associated with a progressive decrease in 10-year survival regardless of the number of involved axillary lymph nodes or size of the primary tumor. Absolute survival differences between high and low grade persisted despite larger tumor sizes and greater nodal involvement, though trends were apparent with increasing breast cancer stage. Statistical significance depended on the number of cases for each combination.

Conclusions.-Histologic grade continues to be of prognostic importance for overall survival despite tumor size and nodal status. Furthermore, these results seem to indicate that the assignment of the histologic grade has been consistent among pathologists when evaluated in a large data set of patients with breast cancer. The incorporation of histologic grade in TNM staging for breast cancer provides important prognostic information.

(Arch Pathol Lab Med. 2014;138:1048-1052; doi: 10.5858/arpa.2013-0435-OA)

The assessment of histologic grade, a composite of tubular differentiation, nuclear features, and mitotic activity, is an important component in the valuation of breast cancers and is a required parameter in the pathologic reporting of breast cancers. It is generally assumed that histologic tumor grade plays an important prognostic role in early-stage cancers with no or few metastatically involved axillary nodes. In contrast, when multiple axillary lymph nodes are involved, the assumption has been made that the contribu- tion of histologic grade is no longer prognostic, and the extent of nodal involvement becomes the primary determi- nant of outcome. While the status of axillary lymph nodes is critical for outcome, we investigated, by using data from the Surveillance, Epidemiology, and End Results Program (SEER) of the National Cancer Institute, whether histologic grade remains a prognostic factor regardless of the size of the primary tumor and/or the number of lymph nodes involved.

MATERIALS AND METHODS

Data Source

Breast cancer patient data were obtained from the SEER Program for the years 1990-2000 by using Case Listing.1 These years were selected to allow for 10 years of follow-up. A total of 273 231 cases of invasive breast carcinoma were obtained. Of the 273 231 cases, 161 708 were graded. To insure uniformity in histologic type, only cases recorded as ''infiltrating ductal carcinoma'' (ICD-O ¼ 8500) and infiltrating ductal comedo type (ICD-O ¼ 8501) were selected and all other histologic types were excluded.2 In situ carcinomas and cases of male breast cancer were also excluded. Cases of grade 4 were combined with cases of grade 3 because the outcome for cases assigned grade 3 or grade 4 is not significantly different.3 Moreover, only 3% of cases were listed as grade 4. Cases with data missing in histologic tumor type, grade, survival time, and vital status were also excluded. The SEER Program is in the public domain. Inflammatory breast cancers and other T4 tumors were not considered in the analysis. Metastatic tumors to other visceral organs and sites were excluded and only M0 cancers were taken into account.

Study Design

We evaluated the 10-year disease-specific survival rate according to the Kaplan-Meier procedure for every combination of T, N, and grade for breast cancer. The T and N variables were categorized according to the American Joint Committee on Cancer.

Calculating Survival for Combinations of Prognostic Factors

Survival rates for every combination of T, N, and grade for breast cancer were calculated by an algorithm programmed in Mathema- tica (Wolfram Research, Champagne, Illinois), which was modeled after the computer program published by Chen et al.4 This algorithm can calculate the individual survival rates for all possible combinations of any number and type of prognostic factors in any order directly from a SEER database. In this study, there were 36 combinations of 3 grade assignments, 3 tumor size groups, and 4 nodal groups (33334), and therefore 36 survival rates. However, since the survival rates were compared, the TN categories were not collapsed into stages, which permitted an analysis of every survival rate. Confidence intervals were also determined but are not reported, since we observed that with large cohorts, the intervals were small and the results usually significant statistically (Figure 1). When the number of cases in specific combinations was small, the confidence intervals were large and often overlapped.

RESULTS

The survival associated with T1N2 breast cancer derived from SEER data recapitulates that identified in the literature and is approximately 68%. This plot may be additionally stratified by including the contribution of histologic grade to the overall survival. Figure 1 shows the 10-year survival rates with confidence intervals for T1N2 breast cancer cases recorded as grades 1, 2, or 3. Similarly, survival rates were generated for all 36 combinations of the histologic grade, tumor size, and nodal status (data not graphed). The Table shows the survival at 10 years for all 36 combinations of grade and TN categories. As expected, within a defined nodal group, the overall survival decreases with increasing tumor size. For a given tumor size, the overall survival decreases with an increase in the number of metastatic lymph nodes. The Table also displays that for every paired combination of T and N, the 10-year survival declined with an increase in grade assignment. Furthermore, the Table also permits a comparison of the effect of increasing the number of positive nodes or increasing tumor size on outcome for each T and N grouping. For instance, for tumors 20 mm in size or smaller and grade 2, the survival for patients with 1 to 3 positive nodes is 0.87 and for patients with 4 to 9 positive nodes it is 0.73. One can also appreciate that grade 2 was most often assigned: 70 477 cases; 65 443 cases were listed as grade 3 and only 25 788 as grade 1. In this manner, the number of cases within a cohort, containing a defined set of T, N and grade variables, correlates with the probability of identifying a subgroup within the total population. Cases assigned T1N2 and grade 3 are more prevalent than those assigned T1N2 and grade 1. Some expected relationships of stage and survival were noted. As expected, the most favorable outcome was seen with small tumors, histologic grades 1 to 3, and no positive nodes. Small tumors assigned grade 1 and associated with 1 to 3 positive nodes also had a favorable outcome. Obviously, if the histologic grade was not prognostic for some combinations of tumor size and nodal status, then the survival rates for those combinations would have remained the same and would not have decreased with increasing grade. Although the data are not shown, similar results on the relation of grade to outcome were obtained when estrogen receptor status, creating 72 combinations, was added to the grade and TN categories.

The data from the Table are depicted in Figure 2, which plots 10-year survival relative to stage categories and their component TN subgroups and stratified according to histologic grade. As 10-year survival decreases with higher tumor stages, the 3 lines are separate and do not merge, which is indicative of 3 distinct cohorts based on the histologic grade. If tumor grade were not a prognostic factor, regardless of tumor size and specific number of nodes, then the survival curves reflecting the levels of grade would overlap or merge at higher stage categories and the curves would no longer be distinguishable.

To further demonstrate the effect of histologic grade on 10-year survival relative to the T and N categories, Figure 3 plots the absolute survival difference of high-grade relative to low-grade tumors versus the nodal groups. The 3 lines are fixed according to tumor size and demonstrate 3 separate curves, which show an initial monotonic increase with nodal status. Interestingly, as tumor size and nodal status increase, the absolute survival difference between low- and high-grade tumors decreases, yet, histologic grade continues to have a significant survival prognosis even at the highest tumor size and nodal group.

COMMENT

Robert B. Greenough,5 in 1925, initially demonstrated the prognostic benefit of the histologic grade for breast cancer. He found 3 factors, namely, tubule formation, nuclear pleomorphism, and hyperchromatism, that correlated with survival. Similar observations followed by others including Patey and Scarff6 in England in 1928. Again, in 1957, Bloom and Richardson7 further showed the prognostic benefit of a grading system based on tubular formation, mitotic index, and nuclear pleomorphism. Modifications of Bloom and Richardson's criteria have further improved the histologic grading of breast cancer.8 The results of a SEER study on the effect of grade on survival also demonstrated that grade independently has a strong effect on the clinical outcome of breast cancer. The study also supported the experience that pathologists generally show concordance in grade assess- ment given the significant differences between the curves.3 Even though a potentially reliable prognostic factor, histologic grade has not been integrated into the TNM. Some have claimed that grade is associated with a low degree of reproducibility9 and a lack of common reporting standards, and that grade provides no guide for the extent of surgery, which is one purpose of the TNM.10

In this study, however, we took a different approach in evaluating the relationship of grade to T and N. Using SEER data, we determined the outcome for each combination of T, N, and grade, but did not collapse the TN combinations into stages. As a result, we were able to work with the individual survival rates of each combination. There were 36 survival rates generated from our algorithm (Table).

As the histologic grade increased from G1 to G3 for each combination of T and N, the survival rates progressively decreased as the tumor size and number of involved nodes increased for each grade level (Figure 2). Consequently, the assumption that the grade is no longer prognostic when large numbers of lymph nodes are involved may be false. Although histologic grade does not necessarily relate to the size of the primary tumor, it may be informative about tumor biology, and hence provide alternative approaches to the therapy of breast cancer in addition to surgery.

The consistent pattern of survival observed between grade, tumor size, and number of involved nodes suggests that pathologists are consistent in the assignment of the histologic grade. Studies have indicated that the assignment of grade is not always reproducible.9 However, when evaluated in a large nonselected data set without collapsing into stage groupings, the results seemed to indicate that pathologists are consistent and that the assignment of grade is reliable.

In addition to the observation that histologic grade maintains its prognostic role for overall survival even at higher tumor sizes and nodal involvement, it is interesting to note that the absolute difference in survival between low and high grade appears to increase with increasing nodal status (Figure 3). This result suggests that at increasing tumor burden (tumor stage), the tumor biologic behavior, using histologic grade as a surrogate marker, differentiates between those relatively aggressive malignancies that show early dissemination from those more indolent tumors that are characterized by a more linear spread.

Finally, the integrity of SEER should be considered within the framework of our conclusions. Critics may argue that SEER does not offer or provide standard criteria for use in the morphologic classification of reported data, but relies upon data submitted from hospital medical records. A pathologic review of the breast cancers reported to SEER, even a representative sample, admittedly would have been ideal, but retrospective access to SEER case material is restricted owing to confidentially and geographic dispersal. Additionally, one may argue that SEER data do not include treatment modalities and prevent comparisons among SEER-reporting sites. However, we suggest that, given the impressive size of the overall population and the narrow confidence intervals based on the large cohort subpopula- tions and their conformity to published results, the data represent a true indication of breast cancer outcome based on the prognostic variable under study. With respect to SEER data on histologic grading in the absence of independent pathologic review, one should note the internal consistency of the data analysis and outcome results. The clear separation of the curves based on histologic grade is consistent given such a large cohort and that interobserver assignment of the histologic grade tends not to be controversial. We note, moreover, that there has been 1 central histologic review of cases submitted to SEER. A fortuitous study on lung cancer conducted in Iowa revealed the reliability of SEER. A special panel of experts reviewed statewide cases of lung cancer because of suspected radon exposure. Since the entire state is a SEER registry, the review covered cases previously submitted to SEER. A comparison showed no statistical difference in the distribution of histologic types reported in the Radon Lung Study when matched with those recorded in the SEER cancer registry.11

In summary, this analysis demonstrates that, for breast cancer, the histologic grade remains a prognostic factor despite changes in tumor size and number of positive lymph nodes. Furthermore, calculating the survival for every combination of prognostic factors has another advantage, for the physician can estimate outcome for patients who have similar prognostic factors without grouping the factors into stages. Although the grade along with other prognostic factors have not been integrated into the TNM, integration is now possible with algorithms that cluster cancer patient data.12,13

References

1. National Cancer Institute, National Institutes of Health. Surveillance Research Program, National Cancer Institute SEER*Stat software version 8.0.2. http://seer.cancer.gov/seerstat. Accessed July 15, 2013.

2. Fritza A, Percy C, Jack A, et al. International Classification of Diseases for Oncology. 3rd ed. Geneva, Switzerland: World Health Organization; 2000:80.

3. Henson DE, Ries L, Freedman L, Carriaga M. Relationship among outcome, stage of disease, and histologic grade for 22,616 cases of breast cancer. Cancer. 1991:68(10):2142-2149.

4. Chen D, Xing K, Henson D, Sheng l, Schwartz AM, Cheng X. Developing prognostic systems of cancer patients by ensemble clustering. J Biomed Biotechnol. 2009;2009:632786. doi:10.1155/2009/632786.

5. Greenough RB, Varying degrees of malignancy in cancer of the breast. J Cancer Res. 1925;9:453-463.

6. Patey DH, Scarff RW. The position of histology in the prognosis of cancer of the breast. Lancet . 1928;1:801-804.

7. Bloom HJG, Richardson WW. Histological grading and prognosis in breast cancer. Br J Cancer. 1957;11:359-377.

8. Elston CW, Ellis IO. Pathological prognostic factors in breast cancer, I: the value of histological grade in breast cancer: experience from a large study with long-term follow-up. Histopathology. 1991;19(5):403-410.

9. Gilchrist KW, Kalish L, Gould VE, et al. Interobserver reproducibility of histopathological features in stage II breast cancer: an ECOG study. Breast Cancer Res Treat. 1985;5(1):3-10.

10. Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, eds. Cancer Staging Manual, American Joint Committee on Cancer.7thed.Geneva, Switzerland: Springer; 2010.

11. Field RW, Smith BJ, Platz CE, et al. Lung cancer histologic type in Surveillance, Epidemiology, and End Results Registry versus independent review. J Natl Cancer Inst . 2004;96(14):1105-1107.

12. Chen LL, Nolan ME, Silverstein MJ, et al. The impact of primary tumor size, lymph node status, and other prognostic factors on the risk of cancer death. Cancer. 2009;115(21):5071-5083.

13. Qi R, Wu D, Sheng L, et al. On an ensemble algorithm for clustering cancer patient data. BMC. In press.

Arnold M. Schwartz, PhD, MD; Donald Earl Henson, MD; Dechang Chen, PhD; Sivasankari Rajamarthandan, BS

Accepted for publication October 22, 2013.

From the Department of Pathology, The George Washington University, Washington, DC (Dr Schwartz); the Division of Cancer Control and Epidemiology, The George Washington University Medical Center Cancer Institute, Washington, DC (Dr Henson and Ms Rajamarthandan); and the Division of Epidemiology and Biostatistics, Uniformed Services University of the Health Sciences, Bethesda, Maryland (Dr Chen).

The authors have no relevant financial interest in the products or companies described in this article.

The opinions expressed in the manuscript are those of the authors and do not necessarily represent those of the Uniformed Services University of the Health Sciences or the Department of Defense.

Reprints: Arnold M. Schwartz, PhD, MD, Department of Pathology, The George Washington University, Kaiser Building, Fifth Floor, 2100-W Pennsylvania Ave, NW Washington, DC 20037 (e-mail: [email protected]).