Although histologic grading of many solid tumors has a long history extending to the early part of this century, 1–6 its practical use has been delayed until recently for breast cancer. The reasons for this delay include the fact that it was not in general use when sophisticated biologically defined elements (such as flow cytometry, oncogenes, kinetic indicators, etc.) and modern clinical trials methodology began in the 1970s. 7–10 Nuclear grading was accepted by some as useful, 11,12 but histologic definitions and testing of interobserver agreement were unavailable. Despite these drawbacks, the 1990 Consensus Conference on Breast Cancer of the National Cancer Institute encouraged pathologists to adopt a uniform grading system. 13
Although many analyses for prognostic indicators settled on size, nodal status, hormone receptors, and sophisticated measures, including cell proliferation, 9,14–17 a preconceived plan of testing a consistently defined, combined histologic grade was begun in Nottingham, England (NCHG). Blamey, Elston and their colleagues 3,18 sought to refine and validate the use of a combined histologic grade that integrates scores from glandular differentiation, nuclear patterns, and mitotic counts. This overall measure is then added to anatomical stage information of tumor size and nodal status to provide powerful prognostic information, rendered in an index validated in several large patient groups. 19,20 The use of the combined histologic grade in routine reporting has been recommended by several major pathology organizations. 21–23
We have used this grade, NCHG, in a large Eastern Cooperative Oncology Group trial of patients with lymph node negative breast cancer, and evaluated interobserver agreement in grade assignment as well as prediction of patient outcome. Because some patients were treated with chemotherapy, the relation of treatment responsiveness to grade could also be evaluated, and compared to S-phase by flow cytometry.
The objectives of this study were to evaluate the reproducibility of a combined histopathologic grading system of breast cancer, to evaluate the ability of the grading system to predict time to treatment relapse and survival, and to use multivariate analyses to evaluate the prognostic importance of the grading data relative to the other clinical and biologic factors.
PATIENTS AND METHODS
Patients for this investigation were taken from EST 1180 (INT0011), which was an Eastern Cooperative Oncology Group coordinated intergroup study of cyclophosphamide/methotrexate/5-fluorouracil/prednisone (CMFP) versus no adjuvant therapy for patients with node-negative breast cancer. High-risk patients [estrogen receptor-negative (ER−) or tumor size ≥3 cm) were randomized between CMFP and no adjuvant therapy, and low-risk patients (<3 cm and ER+) were registered for follow-up. Details of the study have been reported elsewhere. 24–26
In conjunction with EST 1180, blocks were requested, originally for the purpose of performing flow cytometry. 27 The current study, EST 2192 (SWOG-9331), was activated for the purpose of examining the usefulness of histopathologic grading in eligible cases from EST 1180 that were also judged to have adequate tissue for analysis by flow cytometry on an initial screen.
The two pathologists (D.C.A. and D.L.P.) graded the cases independently, without knowledge of the outcome data or other clinical characteristics. Cases were evaluated for slide quality, dominant histologic type, differentiation/growth pattern, mitotic index, and nuclear grade. The last three are the separate elements of the Elston and Nottingham modified Scarff, Bloom, Richardson combined histologic grade system. 3,28 Briefly, differentiation was scored as 3 = poor if less than 10% of the tumor showed definite tubule formation, 2 = moderate for 10% to less than 75%, and 1 = well for more than 75%. Mitotic Index was scored 1 = low if less than 10 mitoses per 10 high-power fields, 2 = medium if 10 to 19, and 3 = high if more than 20. The field diameter for R1 was 0.49 mm. Nuclear grade was scored as 1 = low if there was little variation in the size and shape of the nuclei, 2 = medium for moderate variation, and 3 = high for marked variation and large size. A Combined grade was calculated as the sum of the scores for differentiation, mitotic index, and nuclear grade, with 3 to 5 classified as low grade, 6 to 7 as intermediate grade, and 8 to 9 as high grade.
Also recorded were diagnoses of noninvasive carcinoma only, microinvasive carcinoma, or dominantly noninvasive carcinoma. These categories essentially represent unevaluable cases because they had insufficient invasive carcinoma for evaluation in the material available.
Additional variables considered in this and other analyses of these patients were age at entry (in years), tumor size (pathologic size of the longest diameter, in centimeters), ER status (+ or −), progesterone receptor status (+ or −), ploidy, and percent cells in S phase as defined in Clark et al., 29 HER-2/neu, 17 and p53 expression. 30 Her-2/Neu and p53 were not further analyzed in the current study. The high and low S-phase groups used here were as defined in Clark et al., 29 with low defined as less than or equal to 4.4% for diploid and as less than or equal to 7.0% for aneuploid. Also, some cases with flow cytometry performed were evaluable for ploidy but not S phase.
Time to recurrence (TTR) was measured from date of entry on study to date of first evidence of recurrent or new breast cancer, with follow-up censored at death without recurrence. No effort was made in the clinical database to distinguish metastases of the original tumor from new primary breast cancers, but information was available on involved sites at first recurrence, including opposite breast, so a secondary analysis of the possible impact of new breast cancers on the results could be investigated by censoring follow-up when an opposite breast “recurrence” was observed (these cases were counted as failures in the primary analyses, however). Survival was measured from date of entry on study to date of death from any cause.
The method of Kaplan and Meier 31 was used to estimate TTR and survival curves. The p values comparing groups of curves were from log-rank tests. 32 In various places, both the p value using all follow-up (P), and the p value with follow-up truncated at 5 years (P5), were calculated. For evaluating prognostic effect, the primary interest was in discriminating early failure and death, so only the first 5 years of follow-up were used in determining the significance of effects. This also alleviated concerns about possible nonproportionality of effects during a long follow-up period. The significance of treatment effects within subgroups was determined using all follow-up. Except where noted otherwise, these tests were for marginal associations, ignoring all other factors. Where specifically stated, stratified tests, where contributions to the scores and variances were computed separately within strata, and then summed over strata, were used to test for effects adjusting for the factors used to form the strata.
The partial likelihood method of Cox 33 was used to estimate parameters in proportional hazards regression models. In a particular model, cases missing values of any of the covariates were excluded from the fit. These estimates are asymptotically unbiased, provided the probability that cases were missing covariate values depended only on the covariates, and not on the failure/censoring information. Partial likelihood ratio tests were used to determine the significance of factors in regression models.
In general, associations among patient/disease characteristics are affected by how the sample is drawn. In this study, there were two distinct strata (low and high risk) that were sampled separately, so the pooled data set may not give a representative sample from any well-defined population. Because of this, marginal associations reported here might be different from those that would be observed in specific populations.
There were 927 evaluable patients on E1180, of whom 799 had blocks submitted. Of these, 546 cases were reviewed by both pathologists. Of the 546, 296 cases (54%) were entered in the low-risk stratum (ER+ and tumor size <3 cm) and received no further therapy, and 250 (46%), were entered in the high-risk stratum (ER− and/or tumor size ≥3cm) and were randomized between CMFP and no further therapy; 118 randomized to CMFP and 132 randomized to observation. Of the 118 patients randomized to CMFP, 13 received no adjuvant therapy, whereas of the 132 randomized to no adjuvant therapy, 33 did receive some form of adjuvant therapy. Thus, of the 250 cases in the high-risk stratum, 112 received no adjuvant therapy, whereas 133 received CMFP and 5 received other adjuvant regimens. The results for the two reviewers are referred to by blinded codes R1 and R2.
Median follow-up on the 546 cases was 11.6 years. Of the 546 cases, 167 (31%) had developed breast cancer failures, 27 of which were in the opposite breast, and so might have been new primary breast cancers rather than metastases from the original tumor. Of the 167 failures, all but 40 occurred within 5 years from entry on study, and 9 of the 40 failures after 5 years were in the opposite breast. Of the 546 cases, 154 (28%) had died, 109 after breast cancer treatment failure and 45 without a reported recurrence or new primary breast cancer. Causes of death for these 45 cases were other cancers (12 cases), heart disease (10 cases), pulmonary disease (5 cases), cerebral vascular accident (2 cases), grade V CMFP toxicity (2 cases), and 1 case each from infection, malnutrition, car accident and homicide, and unknown (10 cases).
Overall Slide Quality and Evaluability
Slide quality was judged to be unsatisfactory for 62 (11%) cases by R1 and for 68 (12%) cases by R2. The rate of agreement was excellent, with disagreements for only 16 of 546 (2.9%) of the cases. Excluding the cases judged to be unsatisfactory by either R1 or R2 left 473 cases. Classifying these cases by invasive versus dominant noninvasive versus microinvasive versus noninvasive versus other, the rate of agreement was 461 of 473 (97%). The 48 cases judged by either R1 or R2 to have only noninvasive or microinvasive disease did not have adequate invasive disease on the slides to be graded, and were excluded from the remainder of the analysis, leaving 425 cases. It should be noted that all cases in the study did have invasive disease, regardless of whether it was present on the slides that were reviewed for this project. Cases judged to be of pure medullary type (two cases by both, six by R1 but not R2, five by R2 but not R1) were also excluded, because of their atypical prognosis. One additional case was not graded by R2, leaving 411 cases that were graded by both R1 and R2. Three hundred eleven of these cases received no adjuvant therapy (238 low risk, 73 high risk), and 100 (all high risk) received adjuvant therapy.
Time to Recurrence and Mortality
The analysis is restricted to the 311 cases receiving no adjuvant therapy. Of these cases, 104 had recurred, with 79 of the recurrences in the first 5 years. Thirteen of the recurrences (nine during the first 5 years) were in the opposite breast. Figure 1 shows TTR by NCHG for both reviewers, with p values determined using the first 5 years of follow-up as well as all follow-up. Note that the curves are essentially identical for the two histopathologists. When the analysis was repeated separately for the low-risk and high-risk subgroups, it could be seen that much of the differences in Figure 1 could be attributed to differences in risk groups, and that within risk groups there was much less evidence of differences between histologic grade categories. This was particularly true in the high-risk group that had few patients left untreated. In the low-risk group, however, NCHG continued to have a predictive power (Tables 1 and 2). To further quantify what information that grade adds to tumor size and ER status, regression models were examined considering the factors tumor size (<2.0 versus 2.0–2.9 versus ≥3.0 cm), ER status, risk group, and grade and mitotic index. Tumor size was the most significant factor, and none of the other factors added significantly to the model containing tumor size (Table 1).
Again, the analysis was restricted to the 311 cases receiving no adjuvant therapy. In this subgroup, there had been 93 deaths, of which 30 did not have recurrent disease. Only 31 of these deaths were in the first 5 years, 6 without a prior recurrence. Again, the significance levels and estimated regression effects were determined using only the first 5 years of follow-up.
Both grade and mitotic index alone each showed highly significant association with survival (Fig. 2, Table 2) for both R1 and R2. The similarity of the graphs is evident, and the major importance of mitotic count is also clear in the comparison, because the predictiveness and separation of grades is almost as great for mitotic count alone. As with TTR, the significance was somewhat diminished when tests were stratified on risk group (p = 0.04 for combined grade and p = 0.02 for mitotic index for R1, and p = 0.06 for combined grade and p = 0.009 for mitotic index for R2).
To further examine what grade adds to tumor size and ER status, regression models with the factors risk group, tumor size, and ER status were considered. Risk group was the most significant factor, and neither tumor size nor ER status added significantly to this factor. Table 1 presents the results for grade and mitotic index at 5 years with risk group also included in the model. The predictiveness of both pathologists is seen to be similar.
Predicting Response to Therapy
To evaluate whether the effect of chemotherapy was different within different grading groups, treatment comparisons were made in the subset of high-risk patients evaluable for grade by both R1 and R2, and excluding the cases with medullary histology. This gave 173 patients, of whom 84 were randomized to CMFP and 89 were randomized to no adjuvant therapy. Consistent with other analyses of the effect of therapy for E1180, patients were analyzed as randomized (intention to treat), not as actually treated.
TTR by treatment was highly significant (p = 0.0009) and survival by treatment was not quite significant (p = 0.07) for the full group of 173 patients. Figure 3 shows TTR by treatment for subgroups defined by the combined grade, and Table 3 similarly presents predictiveness of response to treatment by NCHG. Table 3 includes the predictiveness of S-phase, less significant than that of histologic grade.
Many prognostic markers have been tested in the cooperative trials setting 7,12,34–38; however, combined histologic grade has only been tested in single-center trials, however large. 2,39 Both the system of Contesso 2 and that of the Nottingham group 39 have been derived from the Scarff, Bloom, Richardson combined histologic grade 40,41 recommended in the second edition of the World Health Organization book on breast pathology. The Contesso system and the NCHG largely agree when compared, 42 and the NCHG has passed several formal tests of reproducibility among pathologists. 43–45 Recent studies have shown that the prognostic differences seen between young and older women may be accounted for by differences in histologic grade. 46,47 Often when histologic grade is tested with newer modalities such as vascularity growth factors, etc., grade is found to be as strong or stronger. 15,48–50 Our testing of the NCHG demonstrated its usefulness in predicting clinical outcome, particularly survival as well as a further indication that high grade and high mitotic count are predictive of responsiveness to chemotherapy. The two pathologists were in close agreement in assignment of histologic grade, despite completely separate evaluation after a brief meeting to review several cases together.
Table 1 demonstrates the agreement between both observers in terms of outcome prediction for survival, particularly at 5 years, where significance was obtained. Comparison with the tumor size is presented. Figure 1 demonstrates this agreement between pathologists in survival curves, and the TTR and survivals predictability of both observers is shown in Tables 2 and 3. In Figure 2 the fact that much of the predictability depends on the mitotic index is indicated in that the form of the curve with the greatest separation at 5 years is seen to be similar in all displays.
The importance of carefully done grading has been demonstrated in several forms, 16,51 but not in a multicenter cooperative trial. Unfortunately, in this study the histopathologic material was not ideal. This is largely because the tissue blocks were collected retrospectively and had been cut several times for different studies, giving limited material. Particularly, the mitotic index, perhaps being the most important determinant of early death, was not sampled specifically at the outer margins of the tumor, which is the area that gives ideal results. 40,41,52 Certainly grade correlates well with ER positivity, which is one of the reasons why it adds little to that determinant, which was a stratifying variable in this study. However, in some large trials, formal measures of cellular proliferation, closely associated with mitotic counts, have been more powerful predictors than ER. 53 Recently the group that inaugurated this revised approach to grading has shown that the dependence of survival on age is almost completely determined by variations in grade and nodal status between the ages. 47
Because any prognostic determinant finds its greatest clinical use in stratifying patients with regard to therapeutic assignment, it is particularly important to know excellent prognostic groupings 54 as well as the likelihood of responsiveness to therapy. 55 It is this predictiveness of therapeutic responsiveness that has been recently studied with regard to many molecular markers. 17,56 Other studies have suggested that highly proliferative, high grade tumors may have better therapeutic responsiveness to chemotherapy. 49,57–60 The responsivity of tumors in this study was reported in 1989, 25 and this report extends the survival of these patients and indicates a greater responsiveness of high grade cancers to chemotherapy. Figure 3 demonstrates statistical significance for responsiveness to therapy of high NCHG as determined by both pathologists; this significance becomes borderline in the middle group and is absent in the low grade group. This difference is approximately 30%. However, there were few low grade tumors for assessment in comparison because of the assignment of larger ER-negative tumors to the treatment group.
Although this study was hampered by a retrospective design within this multicenter cooperative trial and by relatively small tissue samples, the usefulness of combined histologic grade in determining TTR as well as survival was demonstrated. The lack of ideal material is an important feature of this study; that is, it was successful despite many aspects producing limited and less than ideally prepared materials. Also, two different observers were able to closely agree in assignment of grade and linkage to outcome. Much, but not all, of this information is contained in the mitotic count element of the combined histologic grade; this is particularly strongly demonstrated in survival data as opposed to time to recurrence. Survival and TTR used follow-up only to 5 years for determination of statistical significance. However, survival data for the grades tended to maintain a separation in both survival and TTR after 5 years. However, most data sets demonstrated that many intermediate grade patients had a good survival at 5 years and tended to have more failures at 5–10 years, approximating survivals of high grade patients after 10 years (Fig. 2). This indicates that there is a time dependence to survival by grade, with most of the deaths in patients with high grade tumors appearing early, and a suggestion that deaths in the intermediate NCHG group are fewer and occur later. This time difference as well as overall survival for the three major groups is evident in earlier studies of grade. 2,3
Interest has recently been directed toward predictiveness of responsiveness to certain therapeutic interventions as opposed to prognostic indicators for survival after only local treatment. In this trial, a subset of patients was given chemotherapy and was compared with a group that did not receive such therapy. These patients were in the high-risk category of large tumors and ER− tumors. Within this group, there was a strong indication that high histologic grade predicted better survival after chemotherapy, with a survival difference of approximately 30% at 5 years, which became larger with time, up to more than 10 years. Although the numbers were certainly small, the failures occurred in the untreated group after 5 years, with only a single recurrence in the high combined grade group after 5 years as determined by one of the two pathology observers. There was also a separation in the medium or intermediate grade group, although this did not reach a statistical significance. The numbers in the low grade group were very small but demonstrated no difference between treated and untreated groups.
It is probably especially important to realize that the predictiveness of grade is largely in the first 5 years after diagnosis of breast cancer, but that the separation of survival curves is maintained to 15 years of follow-up. As a determinant of usefulness in assigning patients to chemotherapy, it is interesting that a recent article found that survival in the setting of metastatic disease found that women experiencing early recurrence after initial diagnosis had a particularly poor prognosis. 61 Other studies have shown that patients with earlier recurrences after initial diagnosis had a worse prognosis than those with later recurrence. 62,63 The ability of high grade to predict the women with early recurrence is then understandable when linked to overall survival.
We conclude that combined histologic grade performed conforming to proven guidelines is a robust methodology for identification of characteristics in invasive breast cancers that are not recognized by anatomical staging. Much of this predictive power is derived from mitotic count that is a portion of this system, but the combined grade offers a better interobserver reliability in agreement when rules are followed. 43,44 In this particular grading system, the rules have now been set for more than 10 years and have been tested in many venues. 39–41,64–66 Reference to several studies above have verified the reliability of assignment between pathologists following the few rules necessary. The construction of the Nottingham prognostic index has repeatedly affirmed the addition of prognostic information by the NCHG to available information from anatomical stage of tumor size and nodal status. We expect that the validation of this information will provide a valuable clinical tool in the management of women with breast cancer. 5,20,67,68
1. Baak JPA. Mitosis counting in tumors. Hum Pathol 1990; 21: 683–5.
2. Contesso G, Mouriesse H, Friedman S, et al. The importance of histologic grade in long-term prognosis of breast cancer
: a study of 1,010 patients, uniformly treated at the Institut Gustave-Roussy. J Clin Oncol 1987; 5: 1378–86.
3. 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: 403–10.
4. Miller WR, Ellis IO, Sainsbury J, et al. ABC of breast diseases: prognostic factors. BMJ 1994; 309: 1573–6.
5. Page D, Ellis I, Elston C. Histologic grading
of breast cancer
. Let’s do it [editorial]. Am J Clin Pathol 1995; 103: 123–4.
6. Carriaga MT, Henson DE. The histologic grading
of cancer. Cancer 1995; 75 (suppl 1): 406–21.
7. Mansour EG, Ravdin PM, Dressler L. Prognostic factors in early breast carcinoma. Cancer 1994; 74: 381–400.
8. Clark GM. Integrating prognostic factors. Breast Cancer
Res Treat 1992; 22: 187–91.
9. Dressler LG, Eudey L, Gray R, et al. Prognostic potential of DNA flow cytometry
measurements in node-negative breast cancer
patients: preliminary analysis of an intergroup study (INT 0076). J Natl Cancer Inst Monogr 1992; 11: 167–72.
10. McDivitt RW, Stone KR, Craig RB, et al. A proposed classification of breast cancer
based on kinetic information. Cancer 1986; 57: 269–76.
11. Fisher ER, Costantino J, Fisher B, et al. Pathologic findings from the National Surgical Adjuvant Breast Project (Protocol 4). Discriminants for 15-year survival. National Surgical Adjuvant Breast and Bowel Project Investigators. Cancer 1993; 71: 2141–50.
12. Fisher ER, Redmond C, Fisher B. Prognostic factors in NSABP studies of women with node-negative breast cancer
. National Surgical Adjuvant Breast and Bowel Project. J Natl Cancer Inst Monogr 1992; 11: 151–8.
13. Anonymous. Consensus statement: treatment of early-stage breast cancer
. National Institutes of Health Consensus Development Panel. J Natl Cancer Inst Monogr
14. Silvestrini R, Diadone MG, Di Fronzo G, et al. Prognostic implication of labeling index versus estrogen receptors and tumor size in node-negative breast cancer
. Breast Cancer
Res Treat 1986; 7: 161–9.
15. Axelsson K, Ljung BM, Moore DHN, et al. Tumor angiogenesis as a prognostic assay for invasive ductal breast carcinoma. J Natl Cancer Inst 1995; 87: 997–1008.
16. Wood WC. Integration of risk factors to allow patient selection for adjuvant systemic therapy in lymph node-negative breast cancer
patients. World J Surg 1994; 18: 39–44.
17. Allred DC, Clark GM, Tandon AK, et al. HER-2/neu in node-negative breast cancer
: prognostic significance of overexpression influenced by the presence of in situ carcinoma. J Clin Oncol 1992; 10: 599–605.
18. Le Doussal V, Tubiana-Hulin M, Friedman S, et al. Prognostic value of histologic grade nuclear components of Scarff-Bloom-Richardson (SBR): an improved score modification based on a multivariate analysis of 1262 invasive ductal breast carcinomas. Cancer 1989; 64: 1914–21.
19. Galea MH, Blamey RW, Elston CE, et al. The Nottingham Prognostic Index in primary breast cancer
. Breast Cancer
Res Treat 1992; 22: 207–19.
20. Simpson JF, Page DL. Prognostic value of histopathology in the breast. Semin Oncol 1992; 19: 254–262.
21. Henson DE, Fielding P, Grignon DJ, et al. College of American Pathologists Conference XXVI on clinical relevance of prognostic markers in solid tumors. Arch Pathol Lab Med 1995; 119: 1109–12.
22. Connolly JL, Fechner RE, Kempson RL, et al. Recommendations for the reporting of breast carcinoma, association of directors of anatomic and surgical pathology. Am J Clin Pathol 1995; 104: 614–9.
23. Sloane JP, Anderson TJ, Davies JD, et al. Pathology reporting in breast cancer
screening. 2nd ed. Sheffield, UK:National Health Service Breast Screening Programme,
24. Gray R. Analysis of EST 1180. Technical report no 580E. Boston:Division of Biostatistics, Dana-Farber Cancer Institute,
25. Mansour EG, Gray RJ, Shatila AH, et al. Efficacy of adjuvant chemotherapy in high risk node negative breast cancer
; an intergroup study (NT 0011). N Engl J Med 1989; 320: 485–90.
26. Mansour EG, Eudey L, Shatila AH, et al. Adjuvant therapy of node-negative breast cancer
. Is it necessary for all patients? In: Salmon SE, ed. Adjuvant therapy of cancer VI. Philadelphia:WB Saunders,
27. Gray R. Analysis of EST 7186 (INT 0076): detailed results for the original (TERAK) flow cytometry
data. Technical report no. 714E. Boston:Division of Biostatistics, Dana-Farber Cancer Institute,
28. Elston CW, Ellis IO. Method for grading breast cancer
. J Clin Pathol 1993; 46: 189–90.
29. Clark GM, Mathieu M-C, Owens MA, et al. Prognostic significance of S-phase fraction in good-risk, node-negative breast cancer
patients. J Clin Oncol 1992; 10: 428–32.
30. Elledge RM, Gray R, Mansour E, et al. Accumulation of p53 protein as a possible predictor of response to adjuvant combination chemotherapy with cyclophosphamide, methotrexate, fluorouracil, and prednisone for breast cancer
. J Natl Cancer Inst 1995; 87: 1254–6.
31. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: 457–81.
32. Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 1966; 50: 163–70.
33. Cox DR. Regression models and life-tables (with discussion). J R Stat Soc Ser B 1972; 34: 187–220.
34. Rosen PP, Groshen S, Kinne DW. Survival and prognostic factors in node-negative breast cancer
: results of long-term follow-up studies. J Natl Cancer Inst Monogr 1992; 11: 159–62.
35. Prentice RL. Tumor marker utility grading system. J Natl Cancer Inst 1996; 88: 1419–20.
36. Straus K, Lichter A, Lippman M, et al. Results of the National Cancer Institute early breast cancer
trial. J Natl Cancer Inst Monogr 1992; 11: 27–32.
37. Gelber RD, Goldhirsch A. Reporting and interpreting adjuvant therapy clinical trials. International Breast Cancer
Study Group (formerly Ludwig Group). J Natl Cancer Inst Monogr 1992; 11: 59–69.
38. Mouridsen HT, Andersen J, Andersen KW, et al. Classical prognostic factors in node-negative breast cancer
: the DBCG experience. J Natl Cancer Inst Monogr 1992; 11: 163–6.
39. Todd JH, Dowle C, Williams MR, et al. Confirmation of a prognostic index in primary breast cancer
. Br J Cancer 1987; 56: 489–92.
40. Elston CW. Grading of invasive carcinoma of the breast. In: Page DL, Anderson TJ, eds. Diagnostic histopathology of the breast. Edinburgh, Churchill Livingstone,
41. Elston CW, Ellis IO. Assessment of histological grade. In: Elston CW, Ellis IO, eds. The breast. Edinburgh: Churchill Livingstone, 1998: 365–84.
42. Chouinard EE, Chen V, D’Souza TJ, et al. Reproducibility of pathological grading in breast cancer
. Breast 1994; 3: 124–9.
43. Dalton LW, Page DL, Dupont WD. Histologic grading
of breast carcinoma: a reproducibility study. Cancer 1994; 73: 2765–70.
44. Frierson HF, Wolber R, Berean KW, et al. Interobserver reproducibility of the Nottingham modification of the Bloom and Richardson grading scheme for infiltrating ductal carcinoma. Am J Clin Pathol 1995; 103: 195–8.
45. Robbins P, Pinder S, de Klerk N, et al. Histological grading of breast carcinomas: a study of interobserver agreement. Hum Pathol 1995; 26: 873–9.
46. Pratap R, Shousha S. Breast carcinoma in women under the age of 50: relationship between p53 immunostaining, tumour grade, and axillary lymph node status. Breast Cancer
Res Pract 1998; 49: 35–9.
47. Kollias J, Elston CW, Ellis IO, et al. Early-onset breast cancer
–histopathological and prognostic considerations. Br J Cancer 1997; 75: 1318–23.
48. Barbareschi M, Carlo O, Veronese S, et al. Bcl-2 and p53 expression in node-negative breast carcinoma: a study with long-term follow-up. Hum Pathol 1996; 27: 1149–55.
49. O’Reilly SM, Camplejohn RS, Rubens RD, et al. DNA flow cytometry
and response to preoperative chemotherapy for primary breast cancer
. Eur J Cancer 1992; 28: 681–3.
50. Pinder SE, Wencyk P, Sibbering DM, et al. Assessment of the new proliferation marker MIB1 in breast carcinoma using image analysis: associations with other prognostic factors and survival. Br J Cancer 1995; 71: 146–9.
51. Schumacher M, Schmoor C, Sauerbrei W, et al. The prognostic effect of histological tumor grade in node-negative breast cancer
patients. Breast Cancer
Res Treat 1993; 25: 235–45.
52. Connor AJM, Pinder SE, Elston CW, et al. Intratumoral heterogeneity of proliferation in invasive breast carcinoma evaluated with MIB1 antibody. Breast 1987; 6: 171–6.
53. Bonadonna G, Valagussa P, Tancini G, et al. Current status of Milan adjuvant chemotherapy trials for node-positive and node-negative breast cancer
. Natl Cancer Inst Monogr 1986; 1: 45–9.
54. McGuire WL, Tandon AK, Allred DC, et al. Treatment decisions in axillary node-negative breast cancer
patients. J Natl Cancer Inst Monogr 1992; 11: 173–80.
55. Goldhirsch A, Wood WC, Senn H-J, et al. International consensus panel on the treatment of primary breast cancer
. Eur J Cancer 1995; 31A: 1754–9.
56. MacGrogan G, Bonichon F, De MI, et al. Prognostic value of p53 in breast invasive ductal carcinoma. Breast Cancer
Res Treat 1995; 36: 71–81.
57. Abu-Farsakh H, Sneige N, Arkinson EN, et al. Pathologic predictors of tumor response to preoperative chemotherapy in locally advanced breast carcinoma. Breast J 1995; 1: 96–101.
58. Gamel JW, Meyer JS, Province MA. Proliferative rate by S-phase measurement may affect cure of breast cancer
. Cancer 1995; 76: 1009–18.
59. Amadori D, Volpi A, Maltoni R, et al. Cell proliferation as a predictor of response to chemotherapy in metastatic breast cancer
: a prospective study. Breast Cancer
Res Treat 1997; 43: 7–14.
60. Bonetti A, Zaninelli M, Rodella S, et al. Tumor proliferative activity and response to first-line chemotherapy in advanced breast carcinoma. Breast Cancer
Res Treat 1996; 38: 289–7.
61. Yamamoto N, Watanabe T, Katsumata N, et al. Construction and validation of a practical prognostic index for patients with metastatic breast cancer
. J Clin Oncol 1998; 16: 2401–8.
62. Aberizk WJ, Silver B, Henderson IC, et al. The use of radiotherapy for treatment of isolated locoregional recurrence of breast carcinoma after mastectomy. Cancer 1986; 58: 1214–8.
63. Haffty BG, Fischer D, Beinfield M, et al. Prognosis following local recurrence in the conservatively treated breast cancer
patient. Int J Radiat Oncol Biol Phys 1991; 21: 293–8.
64. Pinder SE, Ellis IO, Elston CW. Prognostic factors in primary breast carcinoma. J Clin Pathol 1995; 48: 981–3.
65. Galea MH, Blamey RW, Elston CE, et al. The Nottingham Prognostic Index in primary breast cancer
. Breast Cancer
Res Treat 1992; 22: 207–19.
66. Genestie C, Zafrani B, Asselain B, et al. Comparison of the prognostic value of Scarff-Bloom-Richardson and Nottingham histological grades in a series of 825 cases of breast cancer
: major importance of the mitotic count as a component of both grading systems. Anticancer Res 1998; 18: 571–6.
67. Page DL, Jensen RA, Simpson JF. Premalignant and malignant disease of the breast: the roles of the pathologist. Mod Pathol 1998; 11: 120–8.
68. Simpson JF, Page DL. Status of breast cancer
prognostication based on histopathologic data. Am J Clin Pathol 1994; 102: S3–8.