Fax: (514) 934-8273
Disruption of the expected positive correlation between breast tumor size and lymph node status in BRCA1-related breast carcinoma†
Article first published online: 8 SEP 2003
Published 2003 American Cancer Society
Volume 98, Issue 8, pages 1569–1577, 15 October 2003
How to Cite
Foulkes, W. D., Metcalfe, K., Hanna, W., Lynch, H. T., Ghadirian, P., Tung, N., Olopade, O., Weber, B., McLennan, J., Olivotto, I. A., Sun, P., Chappuis, P. O., Bégin, L. R., Brunet, J.-S. and Narod, S. A. (2003), Disruption of the expected positive correlation between breast tumor size and lymph node status in BRCA1-related breast carcinoma. Cancer, 98: 1569–1577. doi: 10.1002/cncr.11688
The opinions expressed herein do not necessarily reflect the views of the U.S. Government.
- Issue published online: 3 OCT 2003
- Article first published online: 8 SEP 2003
- Manuscript Accepted: 8 JUL 2003
- Manuscript Revised: 4 JUL 2003
- Manuscript Received: 1 MAY 2003
- Canadian Genetic Diseases Network
- Susan G. Komen Foundation
- Fonds de la recherche en Santé du Québec
- National Institutes of Health
- breast carcinoma;
- tumor size;
- axillary lymph nodes;
- early diagnosis
A positive correlation between breast tumor size and the number of axillary lymph nodes containing tumor is well established. It has been reported that patients with BRCA1-related breast carcinoma are more likely than patients with nonhereditary breast carcinoma to have negative lymph node status. Therefore, the authors questioned whether the known positive correlation between tumor size and lymph node status also was present in women with BRCA1-related breast carcinomas.
The relation between the greatest dimension of the resected breast tumor (size) and the presence of positive axillary lymph nodes (expressed as a percentage of all lymph nodes examined) was evaluated in 1555 women with invasive breast carcinoma who were ascertained at 10 centers in North America between 1975 and 1997. There were 276 BRCA1 mutation carriers, 136 BRCA2 carriers, and 1143 women without a known mutation (208 BRCA1/BRCA2 noncarriers and 935 untested women). Patients were stratified according to tumor size, and odds ratios were estimated for the presence of positive lymph nodes with increasing tumor size.
A highly significant positive correlation between tumor size and the frequency of positive axillary lymph nodes was seen for BRCA1/BRCA2 noncarriers, for BRCA2 carriers, and for untested women (overall P < 0.0001 for each). In contrast, there was no clear correlation between tumor size and positive lymph node status in BRCA1 carriers (overall P = 0.20).
The relation between tumor size and lymph node status in patients with breast carcinoma appears to be different in BRCA1 carriers compared with BRCA2 carriers and noncarriers. These findings have important implications for estimating the route of metastatic spread and for evaluating the effectiveness of early diagnosis in patients with BRCA1-related breast carcinoma. Cancer 2003. ©2003 American Cancer Society.
For most patients with breast carcinoma, tumor size and regional lymph node status are biologic markers of tumor aggressiveness and are independent prognostic factors for survival after diagnosis. Tumor size and the number of positive lymph nodes found at axillary dissection are related directly: a relation that is seen in all large studies performed to date1–6 that have been performed over many decades. Several authors have reported that BRCA1-related breast carcinomas are identified more commonly as axillary lymph node–negative compared with noncarrier controls; however, in general, these studies have been small (for review, see Chappuis et al.7). It has been our impression that tumor size does not bear a strong relation to lymph node status in patients with BRCA1-related breast carcinoma. Therefore, we set out to formally establish whether there is a clear relation between tumor size and lymph node status in BRCA1 mutation carriers (hereafter referred to as carriers). Patients with BRCA2-related breast carcinoma were analyzed separately, because there appear to be important clinical, pathologic, and biologic differences between breast carcinomas arising in BRCA1 carriers and breast carcinomas in BRCA2 carriers.7–10 Of particular relevance to the data presented here is the striking difference between BRCA1 and other subtypes of breast cancer: BRCA1 carriers are more likely than both BRCA2 carriers and noncarriers to develop high-grade and estrogen receptor (ER)-negative tumors.
MATERIALS AND METHODS
Patients Included in Study
Three sets of women with breast carcinoma were studied: Set 1 consisted of a retrospective cohort of 292 Ashkenazi Jewish women who were diagnosed with a first primary invasive breast carcinoma at an age younger than 65 years at 1 Montreal institution between January 1980 and November 1995. They all were tested for the three founder mutations in BRCA1/BRCA2 that are common in the Ashkenazi Jewish population. The tumor size, lymph node involvement, and total number of lymph nodes removed at surgery all were measured and counted by one pathologist (L.R.B.). The ascertainment criteria and mutation analysis performed for these women have been described in previous publications.11, 12 For the current study, all women with tumors > 5.0 cm and women for whom no lymph node dissection was undertaken were excluded. There were 243 eligible women: 29 BRCA1 carriers, 6 BRCA2 carriers, and 208 noncarriers.
Set 2 consisted of a prospective cohort of BRCA1 and BRCA2 carriers who were affected with breast carcinoma. These women were ascertained as part of a North American study of treatment of hereditary breast cancer (Metcalfe et al., unpublished data). There were 10 participating centers. The inclusion criteria were as follows: presence of a germ-line BRCA1 or BRCA2 mutation and invasive breast carcinoma diagnosed in 1975 or thereafter at age 65 years or younger in women resident in North America at the time of diagnosis. Exclusion criteria were: women with a history of breast carcinoma or any other malignancy diagnosed prior to 1975, carcinoma in situ, tumors measuring > 5.0 cm, women with evidence of local or distant metastases at the time of diagnosis, and women with a fixed mass of axillary lymph nodes. Both living and deceased women were included, but the study design precluded survival analysis from the time of diagnosis, because the participants underwent genetic testing at varying times after diagnosis. Data on tumor size and the number of lymph nodes examined (and the number that were positive) were extracted manually from the chart of the patient at the treating institutions. In total, there were 263 BRCA1 carriers and 139 BRCA2 carriers. Five women had mutations in both genes and were counted in both BRCA1 and BRCA2 totals, except when these groups were compared, in which case those five women were excluded. For 16 BRCA1 carriers and 9 BRCA2 carriers, there was incomplete information about the number of lymph nodes that were examined at surgery; therefore, those women were excluded, leaving 247 BRCA1 carriers and 130 BRCA2 carriers in the final analyses from Set 2. When BRCA1 and BRCA2 carriers from Sets 1 and 2 were combined, there were 276 BRCA1 carriers and 136 BRCA2 carriers (412 BRCA1/BRCA2 carriers). For Set 2, patients underwent surgery at many different centers, and there was statistically significant variation in the average number of lymph nodes removed at surgery from center to center (data not shown). It is relevant to the analyses reported here that this difference amounted to an average of 1.2 more lymph nodes examined among BRCA1/BRCA2 carriers compared with noncarriers (P = 0.033). However, because of the relatively small number of BRCA1/BRCA2 carriers at each center, we did not have sufficient statistical power to assess whether the relation between tumor size and lymph node status was the same for all centers.
Set 3 included all women who were diagnosed with invasive breast carcinoma at Women's College Hospital, Toronto, Ontario, Canada between 1987 and 1997. Clinical and histopathologic data were extracted by chart review. There were 1833 women in total; however, for this study, we included only women with tumors that measured ≤ 5 cm, leaving 1513 women. The analysis was restricted further to women for whom complete data on histologic grade, tumor size, number of positive lymph nodes, and total number of lymph nodes examined were available. This resulted in 935 records that were available for the study. The mean number of positive axillary lymph nodes (1.406 lymph nodes) was very similar in the 935 women who were included in the study and the women for whom we had lymph node information (n = 686) but who were not included for other reasons (mean, 1.408 lymph nodes; P = 0.99). None of the women or tumor blocks were tested for BRCA1 or BRCA2 mutations, but the numbers of germ-line BRCA1/BRCA2 mutations were likely to be < 5%; thus, they will not alter the results unduly. This set is referred to as the untested set. For all three sets, tumor size measurements were performed on the macroscopic surgical specimens by pathologists at the respective institutions.
To examine the correlation between tumor size and the frequency of any positive lymph node, we computed an odds ratio by stratifying the tumor size data into 10-mm subgroups, from 0 mm to 50 mm. Individual and overall P values were calculated using the Fisher exact test, and P values for trend were calculated with the Cochran–Armitage test for trend. To compare the relation between the percentage of positive lymph nodes and tumor size, we used an analysis of variance (ANOVA) model in which tumor size was stratified as described above. Heterogeneity of variances between tumor size groupings was assumed. The P value to test an overall effect of tumor size stratification on the percentage of positive lymph nodes was obtained using the F statistic, and pairwise comparisons were deduced from this model. To determine whether the correlation differed by carrier status, all of these statistical analyses were stratified by BRCA1 carriers, BRCA2 carriers, noncarriers, and unknown carriers. We also compared the frequency of positive lymph nodes between BRCA1 carriers, BRCA2 carriers, and noncarriers using the Fisher exact test.
The correlation between tumor size and the percentage of positive lymph nodes was plotted for the three sets of patients with breast carcinoma: Sets 1 and 2 are shown in Figure 1, and Set 3 is shown in Figure 2. In Figure 1, the patients were subdivided by known BRCA1/BRCA2 mutation status; whereas, in Figure 2, tumor grade was used to subdivide Set 3. Each individual was represented by one data point. The percentage of positive lymph nodes was simply the number of lymph nodes that were found to contain tumor by routine pathologic examination divided by the total number of lymph nodes examined at the time of surgery. No special stains were performed to identify micrometastases. The following model was created: percent positive = α + IB1*αB1 + IB2*αB2 + β*tumor size + IB1*βB1*tumor size + IB2*βB2*tumor size, where α is the overall intercept for noncarriers, IB1 = 1 if the sample was composed of BRCA1 carriers (otherwise, it was 0); IB2 = 1 if the sample was composed of BRCA2 carriers (otherwise, it was 0); αB1 is the extra intercept for BRCA1 carriers; αB2 is the extra intercept for BRCA2 carriers; β is the slope for noncarriers; β B1 is the extra slope for BRCA1 carriers; and βB2 is the extra slope for BRCA2 carriers. That is, α represents the part of the intercept that is common to all sets, and β represents the part of the slope that is common to all sets. The same model was used for Set 3, in which the samples were divided into three subgroups according to breast tumor grade (Groups 1, 2, and 3 for women with breast carcinoma Grades 1, 2, and 3, respectively). The terms α and β are the overall intercept and slope; αi and βi are the extra intercept and the extra slope for grade i group (i = 1, 2), respectively. Two-sided P values for each of the variables were calculated.
To explore the association between tumor size and axillary lymph node status, the relation between tumor size and the proportion of women with one or more positive lymph nodes in noncarriers, BRCA1 carriers, and BRCA2 carriers was tabulated, first for women with at least one positive lymph node versus women with negative nodes (Table 1) and then for the average number of positive lymph nodes (Table 2). In BRCA2 carriers, noncarriers, and untested individuals, there was a highly significant, positive association between tumor size and the percentage of patients with positive lymph nodes (overall P < 0.0001; P for trend < 0.0001). In contrast, in breast carcinoma that occurred among BRCA1 carriers, only a weak, statistically borderline correlation between tumor size and lymph node status was seen (overall P = 0.20; P for trend = 0.04). The proportion of lymph node–negative breast carcinomas in BRCA1 carriers changed very little for tumors that measured up to 30 mm in greatest dimension. This is very unlike the situation in BRCA2-related breast carcinomas in noncarriers and in untested individuals. Eight of 47 (17.0%) nonBRCA1/BRCA2-related tumors that measured ≤ 10 mm were lymph node–positive, compared with 10 of 43 tumors (23.2%) among BRCA1 carriers (Table 1) (P = 0.60). In contrast, for tumors measuring 21–30 mm, 25 of 38 (65.8%) non-BRCA1/BRCA2-related breast carcinomas and 16 of 58 (27.1%) BRCA1-related breast carcinomas were lymph node–positive (P = 0.0003).
|Size (mm)||Lymph node status||OR (95% CI)||P value|
|21–30||13||25||9.4 (3.4–25.8)||< 0.0001||—||—|
|BRCA1 mutation carriers|
|BRCA2 mutation carriers|
|31–40||2||11||49.5 (7.3–338.2)||< 0.0001||—||—|
|Total||82||54||—||—||< 0.0001||< 0.0001|
|21–30||119||100||3.10 (1.99–4.83)||< 0.0001||—||—|
|31–40||40||65||6.00 (3.53–10.19)||< 0.0001||—||—|
|41–50||18||32||6.56 (3.34–12.92)||< 0.0001||—||—|
|Total||557||378||—||—||< 0.0001||< 0.0001|
|Size (mm)||No.||Positive lymph nodes (mean ± SD)||P value|
|0–10||7||0.53 ± 2.12||—||—|
|11–20||105||1.81 ± 4.66||0.02||—|
|21–30||38||2.24 ± 2.71||0.002||—|
|31–40||12||4.50 ± 5.04||0.02||—|
|41–50||6||4.67 ± 6.74||0.19||—|
|Total||208||1.84 ± 4.09||—||0.02|
|BRCA1 mutation carriers|
|0–10||43||0.40 ± 0.85||—||—|
|11–20||146||0.71 ± 2.00||0.13||—|
|21–30||58||0.76 ± 1.61||0.15||—|
|31–40||20||2.05 ± 4.05||0.09||—|
|41–50||9||0.67 ± 0.71||0.33||—|
|Total||276||0.77 ± 2.01||—||0.22|
|BRCA2 mutation carriers|
|0–10||30||0.43 ± 1.48||—||—|
|11–20||58||1.19 ± 2.03||0.05||—|
|21–30||28||1.82 ± 2.47||0.01||—|
|31–40||13||5.39 ± 5.32||0.006||—|
|41–50||7||2.00 ± 2.38||0.14||—|
|Total||136||1.60 ± 2.82||—||0.008|
|0–10||121||0.67 ± 2.48||—||—|
|11–20||346||0.75 ± 1.78||0.73||—|
|21–30||247||1.71 ± 3.20||0.0007||—|
|31–40||139||1.97 ± 3.69||0.0009||—|
|41–50||82||3.38 ± 4.14||< 0.0.0001||—|
|Total||935||1.41 ± 2.96||—||< 0.0001|
The mean numbers of positive lymph nodes per size group for BRCA1-related, BRCA2-related, and non-BRCA1/BRCA2-related breast carcinomas also were compared (Table 2). This comparison showed that the mean number of positive lymph nodes per tumor size group also was very different for BRCA1-related breast carcinoma compared with all other groups. It is noteworthy that women with small BRCA1 tumors (0–10 mm) had a mean number of positive lymph nodes (0.40 positive lymph nodes) similar to the number in noncarriers (0.53 positive lymph nodes) and in BRCA2 carriers (0.43 positive lymph nodes) (Table 2); however, this mean number did not increase with increasing tumor size in BRCA1 carriers. The overall P value (ANOVA) for noncarriers was 0.02, but it was 0.22 for BRCA1 carriers. For BRCA2 carriers, the result was similar to the result for noncarriers (P = 0.008) (Table 2). It is noteworthy that the mean number of positive lymph nodes found in association with tumors measuring 21–30 mm was 2.24 positive lymph nodes for noncarriers, 1.82 positive lymph nodes for BRCA2 carriers, and 0.76 positive lymph nodes for BRCA1 carriers (Table 2).
The mean number of positive lymph nodes clearly was influenced by the proportion of tumors that were lymph node–negative. Therefore, we repeated the analyses in Table 2, excluding all patients with negative axillary lymph node status. Because of small numbers in some cells, we stratified the different mutation groups by only two size groups (Table 3). Comparing across groups, BRCA1 carriers with tumors that measured ≤ 30 mm had a mean of 2.39 positive lymph nodes, compared with 4.11 positive lymph nodes for noncarriers (P = 0.007) (Table 3). There were similar but less significant findings in the much smaller group with tumors that measured > 30 mm (Table 3). The significance of these findings was limited by the loss of 60% of the data by excluding patients with negative lymph node status.
|Tumor category||No.||Positive lymph nodes (mean ± SD)||P value|
|All patients (overall P = 0.01)|
|BRCA1 carriers||83||2.55 (2.99)||—||—||0.03||0.05|
|BRCA2 carriers||54||4.02 (3.22)||—||—||—||0.35|
|Tumor size ≤ 30 mm (overall P = 0.06)|
|BRCA1 carriers||69||2.39 (2.65)||—||—||0.15||0.13|
|BRCA2 carriers||38||3.50 (2.19)||—||—||—||0.61|
|Tumor size > 30 mm (overall P = 0.27)|
|BRCA1 carriers||14||3.36 (4.34)||—||—||0.22||0.40|
|BRCA2 carriers||16||5.25 (4.74)||—||—||—||0.45|
In Table 4, the odds ratio (OR) for positive lymph node status according to germ-line BRCA1/BRCA2 mutation status is shown. BRCA1 carriers were significantly less likely to have positive axillary lymph nodes at diagnosis compared with noncarriers (OR, 0.6; P = 0.012). BRCA2 carriers were significantly more likely than BRCA1 carriers to have positive axillary lymph nodes (OR, 1.6; P = 0.043), but BRCA2 carriers did not differ in this respect from noncarriers (OR, 0.9; P = 0.82).
|Mutation status||Lymph node status||OR (95% CI)||P value|
|Noncarriers vs. BRCA1 carriers|
|BRCA1 carriers||193||83||0.6 (0.4–0.9)||0.01|
|Noncarriers vs. BRCA2 carriers|
|BRCA2 carriers||82||54||0.9 (0.6–1.5)||0.82|
|BRCA1 carriers vs. BRCA2 carriers|
|BRCA2 carriers||77||54||1.6 (1.0–2.5)||0.04|
To explore the correlation between tumor size and the percentage of positive lymph nodes further, Sets 1 and 2 (n = 615) were combined, and a linear regression model was used (Fig. 1). A strong positive correlation between tumor size and the number of involved axillary lymph nodes was found (P = 0.0001), as expected. The effect was particularly strong in women without a mutation in BRCA1 or BRCA2 (Fig. 1, dashed blue line) (P = 0.0001). The correlation also was present in BRCA1 carriers (Fig. 1, red line) (P = 0.0039), but the slope was less steep. The BRCA1 slope was statistically significantly different from both the noncarrier slope (P = 0.0001) and the BRCA2 slope (Fig. 1, green line) (P = 0.0027). The BRCA2 slope was intermediate between the BRCA1 and the noncarrier slope and was not significantly different from that seen in noncarriers (P = 0.72). To rule out the possibility that age at diagnosis (younger in BRCA1 carriers) was influencing the results, age of diagnosis was included in the regression model. This made no difference to the results (data not shown). We repeated all the analyses using the actual number of positive lymph nodes, rather than the percentage of the total number of lymph nodes examined, and, again, there was no difference in the results obtained.
To assess whether the behavior of BRCA1-related breast carcinoma with respect to the correlation between tumor size and involvement of axillary lymph nodes was due to other factors, such as tumor histology or histologic grade, we compared the correlation between tumor size and lymph node status in 935 women who were diagnosed with breast carcinoma at a single Toronto hospital between 1987 and 1997 (Set 3). In these women, increasing tumor size had a highly statistically significant association with an increasing probability of positive lymph node status (Table 1). We then compared this correlation separately in women with Grade 1, Grade 2, and Grade 3 tumors (Fig. 2). There was no evidence that tumor grade influenced the correlation (P value for extra slope to account for Grade 2 tumors = 0.59; P = 0.12 for extra slope for Grade 1 tumors) (Fig. 2).
The data from the current study show that the expected association between increasing tumor size and the number of positive axillary lymph nodes is disrupted in patients with BRCA1-related breast carcinoma. Moreover, the results from Tables 1–3 show that the difference in the regression slopes observed between BRCA1 carriers and others (Fig. 1) is not because small BRCA1 tumors are more likely than nonhereditary and BRCA2-related breast carcinoma to be lymph node–positive but, rather, because large BRCA1 tumors are more likely to be lymph node–negative. These findings suggest that basic biologic relations in breast carcinoma differ in BRCA1 carriers from either BRCA2 carriers or noncarriers. Our results mirror the results obtained using detailed histopathologic examination9 and expression or tissue microarrays,13, 14 in which it has been possible to identify a BRCA1-related signature that distinguishes these tumors from tumors in noncarriers13, 14 or in BRCA2 carriers.13 It has been more difficult to distinguish BRCA2-related breast carcinomas from breast carcinomas that occur in noncarriers by using conventional histopathologic measures,9 estrogen receptor or c-erbB-2 (HER2) status,7, 15 or expression microarray data analysis.13 Both tumor size and lymph node status are linked to outcome after breast carcinoma, but they may fail to predict outcome in BRCA1 carriers who develop breast carcinoma. Other indicators of prognosis (that are independent of tumor size and lymph node status) that indicate a propensity for blood-borne metastasis (for example, the over expression of angiogenesis-related markers, such as vascular endothelial growth factor14 or glomeruloid microvascular proliferation16) may be more relevant in BRCA1-related breast carcinoma.
Recently, microarray data have identified a metastasis-specific signature that is common to several tumor types.17 This signature was present in the nondissected primary tumor. Furthermore, a second microarray study identified a prognostically powerful signature that was stronger than, and independent of, tumor size.18 These results imply that when a poor outcome is observed, it is the result of genetic alterations present at a relatively early stage (i.e., before the tumor has enlarged greatly and spread to the regional lymph nodes) and is a property shared by the bulk of the tumor cells present. Obviously, germ-line BRCA1 mutations predate the earliest somatic genetic changes seen in breast carcinoma. Given the data presented here, i.e., that BRCA1-related breast carcinomas tend to be axillary lymph node–negative even if they are large, it may be regarded as surprising that several studies have found that germ-line BRCA1 mutations are often associated with a poor prognosis.8 In particular, women with lymph node–negative, BRCA1-related tumors appear to have a surprisingly poor outcome.12, 19, 20 The results of one of those studies were based on Montreal data presented herein (in Sets 1 and 2). In that study, the breast carcinoma specific survival at 10 years for lymph node–negative BRCA1 carriers was 52%,12 which is worse than what was seen in patients with lymph node–negative tumors of any size in the large SEER study of Carter et al. and equivalent to the survival observed in a woman with a 40–49 mm tumor who had > 4 positive axillary lymph nodes.5 Fewer survival studies have been performed for BRCA2-related breast carcinoma; however, in general, they suggest that the prognosis is not substantially different from that seen in nonhereditary breast carcinoma.8 It is possible that the presence of a BRCA1 mutation increases the chance that the tumor will start out on the wrong foot21 and that one result of this is the uncoupling of the relation between tumor size and lymph node status that we have demonstrated in the current study. In particular, BRCA1-related breast carcinoma may metastasize through mechanisms that differ from the mechanisms that operate in non-BRCA1-related breast carcinoma.
It is not clear why the expected correlation between tumor size and lymph node status does not hold for BRCA1-related breast carcinomas. Recent studies have suggested that BRCA1-related breast carcinomas often present as interval carcinomas22 (i.e., they occur between mammography screenings for breast cancer and are fast-growing tumors). The processes of spread of tumor cells by blood and lymphatic routes may differ: Blood-borne spread is determined by early events in the life of a tumor, whereas lymphatic spread may reflect more closely the time that a tumor has been present.21 If BRCA1-related breast carcinomas grow faster than both BRCA2-related breast carcinomas and nonhereditary breast carcinoma, then they may have a propensity to metastasize to distant sites through the bloodstream rather than through local lymphatic spread. The factors that allow BRCA1-related breast carcinomas to metastasize rapidly and independent of local lymphatic spread also may contribute to the poorer prognosis associated with BRCA1-related breast carcinomas. This fast growth may limit the time available for lymph node metastases to develop. Because we did not observe a similar lack of correlation between tumor size and the percentage of positive lymph nodes in untested women with Grade 3 breast tumors (Fig. 2), it seems unlikely that this feature of BRCA1-related breast carcinomas is due entirely to the fact that most of these tumors will be high-grade lesions. Over expression of proteins such as cyclin E, which promotes unregulated cell cycling and is a predictor of poor prognosis independent of grade,23 may be one such mechanism.
It is possible that our results reflect biases inherent in multiinstitutional retrospective studies, particularly because the subsets analyzed were not derived from a single study with a uniform design but were combined later. We took care, however, to include only those individuals for whom complete data on tumor size, the number of positive axillary lymph nodes, and the total number of lymph nodes examined were available. Observed differences in the mean number of lymph nodes removed per center may affect the mean number of positive lymph nodes, as reported in Table 2; however, it is reassuring that when the linear regression model was applied to the total number of positive lymph nodes rather than the percentage of positive lymph nodes, there was no substantial change in the results obtained. In addition, both BRCA1 and BRCA2 carriers were more likely than noncarriers to have had more lymph nodes examined (an average of 1.2 more lymph nodes). It seems implausible that the statistically significant differences observed between BRCA1 carriers and both BRCA2 carriers and noncarriers with respect to the correlation between tumor size and the percentage of positive lymph nodes (Fig. 1) could be attributable to more lymph nodes being examined in BRCA1/BRCA2 carriers. If the number of lymph nodes removed had been equal in BRCA1 carriers and noncarriers, then the difference in the two slopes likely would be greater than the difference that actually was observed. The number of individuals in the largest tumor size group (41–50 mm) was small: only 4.6% of all patients fell into this size category: Therefore, the estimates of the magnitude of the associations observed are imprecise. The relation between mutation status and the mean number of positive lymph nodes among patients with any positive lymph nodes remains unclear but seems to mirror the results found when patients with negative lymph node status are included. These and other caveats mentioned above indicate that independent replication of our analysis would be welcome, particularly with respect to larger and lymph node–positive tumors.
In conclusion, these findings further confirm the previous observations that BRCA1-related breast carcinomas are biologically distinct from BRCA2-related breast carcinomas, and they also are distinguishable from nonhereditary breast cancer. Our observations may have important implications for programs devoted to the early diagnosis and treatment of women with BRCA1-related breast carcinoma. For example, it will be necessary to evaluate screening trials among BRCA1 carriers using biologic and genomic criteria in addition to size and lymph node status of the detected breast tumors. It also will be valuable to conduct additional epidemiologic studies to determine the prognostic value of tumor size, grade and lymph node status, independently and jointly, in BRCA1 carriers.
The authors thank Dr. J. Ziegler for comments.