©

2003 Blackwell Publishing, Inc., 1075-122X/03/$15.00/0The Breast Journal, Volume 9, Number 3, 2003 167–174

Address correspondence and reprint requests to: Bruce G. Haffty, MD,Department of Therapeutic Radiology, Yale University School of Medicine, P.O.Box 208040, New Haven, CT 06520-8040, USA, or email: bruce.haffty@yale.edu.

Blackwell Science, LtdOxford, UKTBJThe Breast Journal1075-122X2003 Blackwell PublishingMay–June 200393

Original Article

BRCA Status, Molecular Markers, and Clinical Variables

kim et al

.

BRCA

Status, Molecular Markers, and Clinical Variables in Early, Conservatively Managed Breast

Cancer

Sung Kim, MD,* David Rimm, MD, PhD,

†

Darryl Carter, MD,

†

Atif Khan, MD,* Nicole Parisot, BS,* Mayra Alvarez Franco, MD,

‡

Allen Bale, MD,

‡

and Bruce G. Haffty, MD

Departments of

*

Therapeutic Radiology,

†

Pathology, and

‡

Genetics, Yale University School of Medicine, New Haven, Connecticut

�

Abstract:

Conservatively treated premenopausalbreast cancer has a higher rate of local relapse as well as anincreased genetic predisposition to cancer. The current study’spurpose was to evaluate the interactions between

BRCA-1

/

2

status and molecular biologic markers in a cohort of conserv-atively managed breast cancer patients. Seventy-six premen-opausal women treated with breast-conserving surgery andradiation therapy were this study’s focus. All patients weretreated with wide local excision with or without axillary dissec-tion, followed by radiation to the intact breast. Systemic therapywas administered as clinically indicated. All patients in thisstudy had an available paraffin block from the primary tumorand agreed to undergo complete sequencing of the

BRCA-1

and

BRCA-2

genes. The primary breast tumor tissue from eachpatient was immunohistochemically stained for estrogenreceptor (ER), progesterone receptor (PR), p53, HER-2/

neu

,and Proliferating Cell Nuclear Antigen (PCNA). Of the 76patients tested for

BRCA

, 50 patients had wild-type

BRCA-1

and

BRCA-2

, 15 had variants of unclear significance, 6 haddeleterious mutations in

BRCA-1

, and 5 had deleterious muta-tions in

BRCA-2

. p53 positivity correlated with deleteriousmutations in

BRCA-1

(

p

= 0.023), but not in

BRCA-2

. Thoughnot significant, there was a trend for ER and PR negativity tocorrelate with

BRCA-1

mutation (

p

= 0.087 and 0.054, respec-

tively); there were no correlations between ER, PR, and

BRCA-2

. Though not significant, all 11 tumors with

BRCA

mutationswere HER-2/

neu

negative. Patients with

BRCA

mutations havea unique molecular profile. These data can be helpful in under-standing differences in the biologic behavior of patients withfamilial breast cancers.

�

Key Words:

molecular markers

,

BRCA1

,

BRCA2

,

breastcancer

B

reast cancer is an extremely prevalent disease, affect-ing 1 in every 10 women by the age of 70 years (1).

While a

BRCA-1

mutation is associated with an increasedrisk of breast and ovarian cancer, a

BRCA-2

mutation pre-disposes to pancreatic, prostate, hepatic, and male breastcancer as well (2–5). A recent study involving AshkenaziJewish women found that having any of three recurrent

BRCA

mutations conferred a 56% risk of breast cancer bythe age of 70 years. Earlier studies had estimated the riskof breast cancer for a

BRCA-1

carrier to be about 70% byage 70 years and 82% by age 80 years for a

BRCA-2

carrier (6).Since

BRCA-1

and

2

are associated with hereditarycancer and cancer in certain populations with largefounder effects, the subjects of such studies have almostalways been from breast/ovarian cancer families or wereJewish or Icelandic women. In these studies,

BRCA-1

’s

168

•

kim et al

.

correlations with molecular markers has been remarkablyconsistent, while

BRCA-2

’s correlations have been heter-ogeneous. In a study involving Ashkenazi Jewish women,Karp et al. (7) found that

BRCA-1

mutations were asso-ciated with negative estrogen receptor (ER) status. Ina Swedish family study, Johannsson et al. (8) correlated

BRCA-1

mutations to ER, progesterone receptor (PR),and c-

erb

B-2 negativity. In another Scandinavian familystudy, Loman et al. (9) correlated

BRCA-1

mutations tolow ER and PR status, and

BRCA-2

with no significantcorrelations. Noguchi et al. (10), in a Japanese familystudy, found that

BRCA-1

was likely to be more ER andc-

erb

B-2 negative and more p53 positive. They did notfind any correlations involving

BRCA-2

. In a study ofIcelandic breast tumors, Gretarsdottir et al. (11) foundthat

BRCA-2

-positive tumors were p53 positive roughlytwice as often as non-

BRCA-2

tumors, though the resultwas not significant.

In contrast to the numerous studies involving

BRCA

in cancer families and large founder effect populations,there have been very few population-based studies look-ing at

BRCA

mutations and correlations with molecularmarkers. This type of study might shed light on whetherthe mechanisms that cause sporadic breast cancer are thesame as those that cause cancer in families or certain eth-nicities. Armes et al. (12) were the first to conduct sucha population-based study. They chose their subjects onthe basis of having had early breast cancer (before 40 yearsof age). They correlated

BRCA-1

mutations with low ERand c-

erb

B-2 and with high p53 status. On the otherhand,

BRCA-2

had no significant associations. Ames etal. (12) also proposed an intriguing mechanism of how

BRCA-1

interacts with molecular markers. They sug-gested that once

BRCA-1

mutation status is followed byp53 dysfunction, the cell needs no more help to becomecancerous. So there is little selective pressure for cancer cellsto become ER-, PR-, or HER-2/

neu

-positive, hence theunique molecular profile of

BRCA-1

mutant tumors (12).

MATERIALS AND METHODS

Case Selection

One hundred thirty women who had been seen at theDepartment of Therapeutic Radiology, Yale-New HavenHospital between 1975 and 1999 with early premenopau-sal breast cancer were asked to participate in the study.The 76 women who agreed to

BRCA

testing and who hadparaffin-embedded tissue/unstained slides of their primarybreast tumor available for molecular analysis were selectedfor the current study.

Microarray and Unstained Slides

The corresponding paraffin blocks, unstained slides,and hematoxylin and eosin stains were procured eitherfrom Yale-New Haven or other hospitals’ pathologyarchives. The area of cancer in each block was confirmedby a pathologist. The array involved taking a small coresample from within the area of confirmed cancer frommany blocks and then inserting these samples into anothersingle paraffin block, so that confirmed cancerous sam-ples from many different women’s tumors were arrangedin a single block (Fig. 1). The microarray block was thencut for staining as outlined below.

BRCA

Testing

For each patient in the study, blood samples werecompletely sequenced for germline mutations in

BRCA-1

and

BRCA-2

. Leukocyte DNA was directly sequenced usingthe ABI375 automatic sequencer. Mutations causing apremature stop codon in

BRCA-1

or

2

or missensemutations that are known to cause phenotypic cancer weretermed

BRCA

mutations, while missence mutations thatdo not yet have enough collected data on them to call themeither

BRCA

mutations or polymorphisms were termedvariations of unclear significance. Missense mutationsthat have been shown not to cause phenotypic cancer weretermed polymorphisms.

Immunohistochemical Staining

All the tumor samples in the tissue microarray werestained for ER, PR, p53, HER-2/

neu

, and PCNA. Thestaining was performed on 5

µ

m sections from paraffinblocks. The sections were dried at 60

°

C for 60 minutesThe sections were dewaxed in xylene and rehydratedthrough graded alcohols to distilled water. Antigenretrieval was performed for all the immunostains listed in

Figure 1. Tissue microarray for biopsy samples of chest wall relapses.

BRCA Status, Molecular Markers, and Clinical Variables •

169

Table 1. The antigen retrieval method used was citratebuffer pH 6.7 in a steam bath at 97

°

C for 20 minutes.Yale immunohistology has found that this methodworks better than microwaving. The Dako autostainerwas used to perform the antibody staining with thespecific antibodies listed in Table 1. All the wash stepsused

tris

-buffered saline. Sections were treated withhydrogen peroxide for 5 minutes. Since the Dako lsab+and Envision detection kits were used, no blocking stepswere required. The sections were incubated in antibodyat the dilutions listed in Table 1 for 30 minutes at roomtemperature. Secondary antigen retrieval was performedusing the detection kits listed in Table 1. Labeling wasperformed with DAB. All sections were counterstainedwith hematoxylin, dehydrated with ethanol, and perma-nently coverslipped.

Analysis

The pathologists, who were blinded to

BRCA

resultsand clinical data, subsequently read the stained slides.They graded ER and PR from 0 to 300, with a score of 74or higher being positive and 73 or lower being negative.p53, HER-2/

neu

, and PCNA were scored from 0 to 3 witha score of 2 or 3 recorded as positive and 0 or 1 as negative.Univariate analysis was performed using PRODAS statis-tical software to find correlations between

BRCA

statusand ER, PR, HER-2/

neu

, p53, PCNA status and clinicaldata including pathologic tumor stage, pathologic nodalstatus, tumor margin status, method of tumor detection,type of histology, adjuvant chemotherapy use, adjuvanthormone therapy use, and family history.

RESULTS

The median age of the 76 study subjects was 39 years.Median follow-up was 10.4 years. Of the subjects withavailable family history, 33 had a positive family historyand 41 did not. Of the 76 study subjects, 39 received adju-vant chemotherapy and 15 received hormone therapy.Eighteen had a breast recurrence, 8 had metastases, 1 hada nodal recurrence, and 15 had a second malignancy. Atthe time of the analysis, 2 of the subjects had died, 7 wereliving with breast cancer, and 67 were living with noevidence of disease. Due to the selection bias of this study,all patients had to be alive to participate, therefore noattempt was made at that time to correlate the molecularand genetic profiles with outcomes.

BRCA

Testing

Seventy-six tumor samples were analyzed for

BRCA

mutations and molecular markers. There were 11

BRCA

mutations (6

BRCA-1

and 5

BRCA-2

) and 15 variants ofunclear significance (8

BRCA-1

and 7

BRCA-2

) (Table 2).

Molecular Markers

Fifty-five percent of the tumors were ER positive and59% were PR positive. A high percentage (79%) stainedpositively for PCNA. Thirty-three percent of tumorsstained positively for p53 and 8% for HER-2/

neu

(Table 3).Correlation of the molecular markers, BRCA status andclinical variables are summaried in Tables 4–10.

Though not statistically significant,

BRCA-1

tended tocorrelate with ER negativity. Only 1 of 6

BRCA-1

tumors(17%) were ER positive compared to 4 of 5

BRCA-2

Table 1. Immunostaining Data

Antibody Dilution Source Catalog no. AG retrieval Detection kit

ER 1:2 Dako N1575 Steam EnvisionPR 1:2 Dako N1630 Steam EnvisionHER-2/neu Prediluted Dako Hercep K5204p53 1:400 Dako M7001 Steam Dako lsab+PCNA 1:6400 Dako M0879 Steam Dako lsab+

Table 2. BRCA Testing Results

Total number tested for BRCA mutations 76BRCA-1 mutation 6BRCA-2 mutation 5BRCA-1 variants of unclear significance 8BRCA-2 variants of unclear significance 7Wild type or polymorphism 50

Table 3. Immunostaining Results

ER positive 40/73 (55%)PR positive 44/75 (59%)HER-2/neu positive 6/75 (8%)p53 positive 25/75 (33%)PCNA positive 53/67 (79%)

170

•

kim et al.

tumors (80%) and 35 of 62 non-BRCA tumors (56%)(p = 0.087); 1 of 6 BRCA-1 tumors (17%) and 2 of 5BRCA-2 tumors (40%) were PR positive versus 41 of 64non-BRCA tumors (64%) (p = 0.054). Collectively 3 of

11 BRCA-positive tumors (27%) (both BRCA-1 and 2)were PR positive compared to 41 of 64 non-BRCA tumors(64%) (p = 0.022). BRCA-1 also correlated with p53positivity: 5 of 6 BRCA-1 tumors (83%) were p53 positive

Table 4. Immunohistochemical and Clinical Variables versus BRCA Status (BRCA-1 and 2 Mutations TakenCollectively)

Variable BRCA mutation (1 or 2) No BRCA mutation p

ER positive 5/11 (45%) 35/62 (56%) 0.499PR positive 3/11 (27%) 41/64 (64%) 0.022p53 positive 6/11 (55%) 19/64 (30%) 0.106HER-2/neu positive 0/11 (0%) 6/64 (9%) 0.290PCNA positive 9/11 (82%) 44/56 (79%) 0.809Family history 8/10 (80%) 25/64 (39%) 0.015Adjuvant chemotherapy 7/11 (64%) 32/65 (49%) 0.377Adjuvant hormone therapy 2/11 (18%) 13/65 (20%) 0.889Stage 0.452

T1 9/11 (82%) 47/65 (72%)T2 2/11 (18%) 18/65 (28%)

N stageN0 7/10 (70%) 47/57 (82%) 0.358N1 3/10 (30%) 10/57 (18%)

Table 5. Molecular Profile and Clinical Variables versus BRCA Status

Variable BRCA-1 mutation BRCA-2 mutation No BRCA mutation p

ER positive 1/6 (17%) 4/5 (80%) 35/62 (56%) 0.087PR positive 1/6 (17%) 2/5 (40%) 41/64 (64%) 0.054p53 positive 5/6 (83%) 1/5 (20%) 19/64 (30%) 0.023HER-2/neu positive 0/6 (0%) 0/5 (0%) 6/64 (9%) 0.571PCNA positive 6/6 (100%) 3/5 (60%) 44/56 (79%) 0.259Family history 5/5 (100%) 3/5 (60%) 25/64 (39%) 0.024Adjuvant chemotherapy 4/6 (67%) 3/5 (60%) 32/65 (49%) 0.660Adjuvant hormone therapy 2/6 (33%) 0/5 (0%) 13/65 (20%) 0.381T stage 0.809

T1 5/6 (83%) 4/5 (80%) 47/65 (72%)T2 1/6 (17%) 1/5 (20%) 18/65 (28%)

N stageN0 4/6 (67%) 3/4 (75%) 47/57 (82%) 0.621N1 2/6 (33%) 1/4 (25%) 10/57 (18%)

Table 6. Variable versus ER Status

Variable ER positive/negative ER negative p

PR positive 35/40 (88%) 9/33 (27%) 0.000p53 positive 9/40 (23%) 16/33 (48%) 0.020HER-2/neu positive 3/40 (8%) 3/33 (9%) 0.805PCNA positive 28/36 (78%) 23/28 (82%) 0.667Adjuvant chemotherapy 16/40 (40%) 22/33 (67%) 0.023Adjuvant hormone therapy 8/40 (20%) 6/33 (18%) 0.844Family history 14/39 (36%) 17/32 (53%) 0.145Stage 0.188

T1 33/40 (82%) 21/33 (64%)T2 7/40 (18%) 12/33 (36%)

N stageN0 26/32 (81%) 25/32 (78%) 0.756N1 6/32 (19%) 7/32 (22%)

BRCA Status, Molecular Markers, and Clinical Variables • 171

Table 8. Immunohistochemical and Clinical Variables versus p53

Variable p53 positive/negative p53 negative p

ER positive 9/25 (36%) 31/48 (65%) 0.020PR positive 10/25 (40%) 34/50 (68%) 0.020HER-2/neu 3/25 (12%) 3/50 (6%) 0.367PCNA 20/24 (83%) 32/42 (76%) 0.495BRCA-1 5/25 (20%) 1/50 (2%) 0.007BRCA-2 1/25 (4%) 4/50 (8%) 0.513Adjuvant chemotherapy /25 (68%) 22/50 (44%) 0.050Adjuvant hormone therapy 7/25 (28%) 8/50 (16%) 0.220Family history 16/25 (64%) 16/48 (33%) 0.012T stage 0.805

T1 18/25 (72%) 37/50 (74%)T2 7/25 (28%) 13/50 (26%)

N stageN0 16/23 (70%) 37/43 (86%) 0.109N1 7/23 (30%) 6/43 (14%)

Table 7. Immunohistochemical and Clinical Variables versus PR Status

Variable PR positive/negative PR negative p

ER 35/44 (80%) 5/29 (17%) 0.000HER-2/neu 3/44 (7%) 3/31 (10%) 0.653p53 10/44 (23%) 15/31 (48%) 0.020PCNA 30/37 (81%) 22/29 (76%) 0.607BRCA (1 or 2) 3/44 (7%) 8/31 (26%) 0.022BRCA-1 positive 1/44 (2%) 5/31 (16%) 0.029BRCA-2 positive 2/44 (5%) 3/31 (10%) 0.380Adjuvant chemotherapy 19/44 (43%) 20/31 (65%) 0.069Adjuvant hormone therapy 9/44 (20%) 6/31 (19%) 0.910Family history 14/43 (33%) 18/30 (60%) 0.020T stage 0.577

T1 32/44 (73%) 23/31 (74%)T2 12/44 (27%) 8/31 (26%)

N stageN0 29/37 (78%) 24/29 (83%) 0.657N1 8/37 (22%) 5/29 (17%)

Table 9. Immunohistochemical and Clinical Variables versus HER-2/neu

Variable HER-2/neu positive/negative HER-2/neu negative p

ER positive 3/6 (50%) 37/67 (55%) 0.805PR positive 3/6 (50%) 41/69 (59%) 0.653p53 3/6 (50%) 22/69 (32%) 0.367PCNA 4/5 (80%) 48/61 (79%) 0.945BRCA-1 0/6 (0%) 6/69 (9%) 0.451BRCA-2 0/6 (0%) 5/69 (7%) 0.494Adjuvant chemotherapy 4/6 (67%) 35/69 (51%) 0.453Adjuvant hormone therapy 4/6 (67%) 11/69 (16%) 0.003Family history 2/6 (33%) 30/67 (45%) 0.588T stage 0.543

T1 5/6 (73%) 50/69 (72%)T2 1/6 (17%) 19/69 (28%)

N stageN0 4/5 (80%) 49/61 (80%) 0.986N1 1/5 (20%) 12/61 (20%)

172 • kim et al.

compared to 1 of 5 BRCA-2 tumors (20%) and 19 of 64non-BRCA tumors (30%) (p = 0.023). Though the rela-tionship between BRCA mutation status and HER-2/neunegativity was not statistically significant, it was notablethat all 11 BRCA-positive tumors were HER-2/neu nega-tive compared with 6 of 64 of the non-BRCA tumors (9%)(p = 0.571). Predictably BRCA mutation status correlatedto a positive family history of breast cancer: 5 of 5 of theBRCA-1 tumors (100%) and 3 of 5 of the BRCA-2 tumors(60%), as opposed to 25 of 64 of the non-BRCA tumors(39%), had a positive family history (p = 0.024).

Also predictably, positive ER and PR status correlatedwith each other (p = 0.000). ER negativity also correlatedto p53 positivity: 16 of 33 ER-negative tumors (48%)versus 9 of 40 ER-positive tumors (23%) stained for p53(p = 0.020). Of interest is that PR negativity correlatedto positive family history: 18 of 30 PR-negative tumors(60%) had a positive family history as opposed to 14 of43 PR-positive tumors (33%) (p = 0.020).

Although the numbers were too small to reach statisti-cal significance, it is notable that none of the patients withBRCA-1 or BRCA-2 mutations had an overexpression ofHER-2/neu.

p53 was involved in several significant associations: 17of 25 p53-positive tumors (68%) received adjuvant chem-otherapy, as opposed to 22 of 50 p53-negative tumors(44%) (p = 0.050). Surprisingly, none of the 25 p53-positive tumors had metastases compared to 8 of 50 p53-negative tumors (16%) (p = 0.034). Since several of thep53-positive tumors were in patients with BRCA mutations,it is possible that p53 positivity in BRCA patients does nothave the same prognostic significance as in patients with

sporadic cancer. p53 also correlated with a positive familyhistory: 16 of 25 of p53-positive tumors (64%) had pos-itive family history, as opposed to 16 of 48 p53-negativetumors (33%) (p = 0.012). This also may be related to thefact that patients with BRCA mutations, who clearly havea positive family history, are also likely to be p53 positive.

DISCUSSION

In this study we took tumor samples from 76 womenwho were chosen on the basis of having had early stagepremenopausal breast cancer, agreeing to BRCA testing,and having available tumor samples. All had receivedconservative surgery and radiation therapy. Correlationswere sought between BRCA status, immunohistochemicalstaining data, and clinical variables including pathologictumor and node status, tumor margin status, method oftumor detection, tumor histology, adjuvant chemother-apy use, adjuvant hormone therapy use, and family his-tory. In this study we intentionally did not look at survivalas a clinical variable since the nature of our protocol (requir-ing active participation and informed consent of patientswith early breast cancer) introduced an unavoidable survivalbias in that any deceased patients were excluded.

For the most part, the correlation that we foundbetween BRCA status and molecular markers was similarto that reported previously in the literature. Thoughnot statistically significant, there was a definite trend forBRCA-1 to be associated with ER negativity (p = 0.087).BRCA-1 had an association with PR negativity that wasnearly significant (p = 0.054). Taken collectively, BRCA-1 and BRCA-2 did have a significant association with PR

Table 10. Immunohistochemical and Clinical Variables versus PCNA

Variable PCNA positive/negative PCNA negative p

ER positive 28/51 (55%) 8/13 (62%) 0.667PR positive 30/52 (58%) 7/14 (50%) 0.607p53 20/52 (38%) 4/14 (29%) 0.495HER-2/neu 4/52 (8%) 1/14 (7%) 0.945BRCA-1 6/53 (11%) 0/14 (0%) 0.187BRCA-2 3/53 (6%) 2/14 (14%) 0.275Adjuvant chemotherapy 33/53 (62%) 5/14 (36%) 0.075Adjuvant hormone therapy 12/53 (23%) 3/14 (21%) 0.922Family history 24/51 (47%) 6/14 (43%) 0.780T stage 0.690

T1 37/53 (70%) 11/14 (79%)T2 16/53 (30%) 3/14 (21%)

N stageN0 39/50 (78%) 8/10 (80%) 0.889N1 11/50 (22%) 2/10 (20%)

BRCA Status, Molecular Markers, and Clinical Variables • 173

negativity (p = 0.022). Also consistent with previous liter-ature was the association of BRCA-1 with p53 positivity(p = 0.023). Statistically there was no significant associa-tion between BRCA status and HER-2/neu, but notably,none of the 11 BRCA-positive tumors were HER-2/neupositive, compared with 6 of 64 of the non-BRCA tumors(9%) (p = 0.290). As anticipated, ER and PR statuswere highly correlated with each other (p = 0.000). BothBRCA-1 and 2 were associated with a positive familyhistory (p = 0.015). Though the number of BRCA-positivepatients in our study was small, several correlationswith molecular markers were still statistically significant,and others showed trends but may have lacked thepower to be significant. These results show that even in apopulation-based study, BRCA-positive tumors have aunique molecular profile, reinforcing the notion that, atsome level, the various BRCA-1 mutations share a com-mon mechanism of tumorigenesis.

In a recently published study we reported a higher rateof ipsilateral and contralateral breast cancer events inpatients with BRCA mutations (13). The current study iscomprised of a subset of patients from that study in whomparaffin material was available. In this subset, the rela-tionship between BRCA status and ipsilateral and contral-ateral events was also noted, but statistical significanceis not as strong due to the smaller patient numbers in thecurrent analysis.

A somewhat puzzling correlation was the one betweenp53 negativity and metastasis (p = 0.034). The relativelysmall numbers and selection bias in our specific patientpopulation may explain this. Though not significant, p53positivity did correlate to higher N status: 30% N1 versus14% N1 (p = 0.109). This correlation of p53 negativityand metastasis is contrary to Overgaard et al.’s (14) resultin a study involving women with early sporadic breastcancer that p53 positivity, not negativity, correlated tometastasis. Overgaard et al.’s study involved both node-negative and node-positive women, the latter receivingadjuvant chemotherapy. Silvestrini et al. (15) similarlyfound that high p53 levels corresponded to distant metas-tases in a study involving 1400 node-negative breastcancers. In our current study, several of the positive p53patients also had BRCA mutations. Whether overexpres-sion of p53 in BRCA carriers maintains the same pro-gnostic significance as in sporadic cancers is unclear. Theeffect of p53 mutations on metastasis and survival in bothsporadic and genetically linked breast cancer warrantsfurther investigation.

The correlations of BRCA-1 with markers such asER, PR, HER-2/neu, and p53 have been remarkably

consistent in studies involving cancer families, certain ethnicpopulations, and in this population-based study as well.This consistency hints that at some level there is a commonmechanism through which BRCA-1 mutations causebreast cancer. In this specific population of premenopau-sal breast cancer patients, ER positivity and PCNA nega-tivity were highly correlated with local recurrence. Theserelationships clearly warrant further investigation in stud-ies involving more patients and taking into account theeffects of node status, margin status, and the use ofsystemic adjuvant therapy.

Acknowledgments

This work was funded by the Ethel F. DonaghueWomen’s Health Investigator Program and the PattersonTrust Fund, Yale University School of Medicine.

REFERENCES

1. Greenlee RT, Murray T, Bolden S, Wingo PA. Cancerstatistics, 2000. CA Cancer J Clin 2000;50:7–33.

2. Cancer risks in BRCA 2 mutation carriers. The BreastCancer Linkage Consortium. J Natl Cancer Inst 1999;91:1310–16.

3. Bertwistle D, Ashworth A. BRCA1 and BRCA2. CurrBiol 2000;10:R582.

4. Bishop DT. BRCA1 and BRCA2 and breast cancerincidence: a review. Ann Oncol 1999;10(suppl 6):113–19.

5. Easton DF, Steele L, Fields P, et al. Cancer risks in twolarge breast cancer families linked to BRCA2 on chromosome13q12-13. Am J Hum Genet 1997;61:120–28.

6. Stratton MR. Recent advances in understanding ofgenetic susceptibility to breast cancer. Hum Mol Genet 1996;5Spec No:1515–19.

7. Karp SE, Tonin PN, Begin LR, et al. Influence of BRCA1mutations on nuclear grade and estrogen receptor status ofbreast carcinoma in Ashkenazi Jewish women. Cancer1997;80:435–41.

8. Johannsson OT, Idvall I, Anderson C, et al. Tumourbiological features of BRCA1-induced breast and ovariancancer. Eur J Cancer 1997;33:362–71.

9. Loman N, Johannsson O, Bendahl PO, Borg A, Ferno M,Olsson H. Steroid receptors in hereditary breast carcinomasassociated with BRCA1 or BRCA2 mutations or unknownsusceptibility genes. Cancer 1998;83:310–19.

10. Noguchi S, Kasugai T, Miki Y, Fukutomi T, Emi M,Nomizu T. Clinicopathologic analysis of BRCA1- or BRCA2-associated hereditary breast carcinoma in Japanese women.Cancer 1999;85:2200–205.

11. Gretarsdottir S, Thorlacius S, Valgardsdottir R, et al.BRCA2 and p53 mutations in primary breast cancer in relationto genetic instability. Cancer Res 1998;58:859–62.

12. Armes JE, Trute L, White D, et al. Distinct molecular

174 • kim et al.

pathogeneses of early-onset breast cancers in BRCA1 andBRCA2 mutation carriers: a population-based study. CancerRes 1999;59:2011–17.

13. Haffty BG, Harrold E, Khan AJ, et al. Outcome ofconservatively managed early-onset breast cancer by BRCA1/2status. Lancet 2002;359:1471–77.

14. Overgaard J, Yilmaz M, Guldberg P, Hansen LL,

Alsner J. TP53 mutation is an independent prognostic markerfor poor outcome in both node-negative and node-positive breastcancer. Acta Oncol 2000;39:327–33.

15. Silvestrini R, Daidone MG, Benini E, et al. Validation ofp53 accumulation as a predictor of distant metastasis at 10 yearsof follow-up in 1400 node-negative breast cancers. Clin CancerRes 1996;2:2007–13.