91 day challenge (preview)

RAPID PUBLICATION

Effect of Preoperative Chemotherapy on Local-RegionalDisease in Women With Operable Breast Cancer:

Findings From National Surgical Adjuvant Breast andBowel Project B-18

By Bernard Fisher, Ann Brown, Eleftherios Mamounas, Samuel Wieand, Andre Robidoux, Richard G. Margolese,Anatolia B. Cruz, Jr, Edwin R. Fisher, D. Lawrence Wickerham, Norman Wolmark, Arthur DeCillis, James L. Hoehn,

Alan W. Lees, and Nikolay V. Dimitrov

Purpose: To determine whether preoperative doxo-ru id ncd cyclophosphamide (AC) permits more lump-ectomies to be performed and decreases the incidence ofpositive nodes in women with primary breast cancer.

Patients and Methods: Women (n = 1,523) were ran-domized to National Surgical Adjuvant Breast and BowelProject (NSABP) B-18; 759 eligible patients receivedpostoperative AC and 747, preoperative AC. The clinicalsize of breast and axillary tumors was determined be-fore each of four cycles of AC and before surgery. Tumorresponse to preoperative therapy was clinically complete(cCR), partial (cPR), stable (cSD), or progressive disease(cPD). Tissue from patients with a cCR was evaluated fora pathologic complete response (pCR).

Results: Breast tumor size was reduced in 80% of pa-tients after preoperative therapy; 36% had a cCR. Tumorsize and clinical nodal status were independent pre-dictors of cCR. Twenty-six percent of women with a cCRhad a pCR. Clinical nodal response occurred in 89% ofnode-positive patients: 73% had a cCR and 44% of those

FOR MOST OF THIS CENTURY, breast cancer man-agement was governed by a paradigm that espoused

the use of expansive surgery and radiation therapy. How-ever, in the 1950s and 1960s, systemic chemotherapy andhormonal therapy began to be used for the treatment ofadvanced disease. With the advent of postoperative adju-vant chemotherapy for earlier stage disease in the mid-1970s, two independent paradigms subsequently arose:one directed toward controlling local-regional disease andthe other aimed at eliminating micrometastatic tumor.Because more effective adjuvant therapy has resulted inbetter control of systemic disease, as well as a reductionin the incidence of breast tumor recurrence after lumpec-tomy, these two independent paradigms have convergedinto a single exemplar.

As these changes were evolving, information from avariety of sources generated interest in the use of preoper-ative (also known as neoadjuvant, primary, or induction)chemotherapy for the treatment of operable breast cancer.In a recent editorial we stated that, although it has beenshown that preoperative chemotherapy results in a markeddecrease in the size of primary breast cancers and that aputative increase in breast-conserving operations may oc-cur as a consequence, information should be obtained

had a pCR. There was a 37% increase in the incidenceof pathologically negative nodes. Before randomization,lumpectomy was proposed for 86% of women with tu-mors : 2 cm, 70% with tumors 2.1 to 5.0 cm, and 3%with tumors ý 5.1 cm. Clinical tumor size and nodal sta-tus influenced the physician's decision. Overall, 12%more lumpectomies were performed in the preoperativegroup; in women with tumors _ 5.1 cm, there was a175% increase.

Conclusion: Preoperative therapy reduced the size ofmost breast tumors and decreased the incidence of posi-tive nodes. The greatest increase in lumpectomy afterpreoperative therapy occurred in women with tumors ý5 cm, since women with tumors less than 5 cm werealready lumpectomy candidates. Preoperative therapyshould be considered for the initial management ofbreast tumors judged too large for lumpectomy.

J Clin Oncol 15:2483-2493. © 1997 by American So-ciety of Clinical Oncology.

from large randomized trials before that therapeutic ap-proach is universally adopted.' Findings from such studiesare likely to promote a better understanding of the bio-logic and clinical consequences of such therapy and, thus,permit its role to be more clearly defined.

Those considerations, and hypotheses formulated frombiologic"4 and clinical information obtained after preoper-

From the Department of Biostatistics, Graduate School of PublicHealth, and the Department of Surgery, University of Pittsburgh,Shadyside Hospital, and Allegheny General Hospital, Pittsburgh,PA; Mt. Sinai Center for Breast Health, Cleveland, OH; St. LucHospital, and Jewish General Hospital, Montreal, and Cross CancerInstitute, Edmonton, Canada; University of Texas, San Antonio, TX;Marshfield Clinic, Marshfield, WI; and Michigan State University,East Lansing, MI.

Submitted October 15, 1996; accepted April 2, 1997.Supported by Public Health Service grants (NCI-UIO-CA 12027,

NCI-U10-CA-37377, and NCI-UlO-CA-39086) from the NationalCancer Institute, Bethesda, MD, and by a grant (ACS-R-13) fromthe American Cancer Society, Chicago, IL.

Address reprint requests to Bernard Fisher, MD, 3550 TerraceSt, Room 914, Pittsburgh, PA 15261; Email [email protected].

C 1997 by American Society of Clinical Oncology.0732-183X/97/1507-0010$3.00/0

Journal of Clinical Oncology, Vol 15, No 7 (July), 1997: pp 2483-2493 2483

Information downloaded from jco.ascopubs.org and provided by at ASCO on May 24, 2011 from 206.205.123.242Copyright © 1997 American Society of Clinical Oncology. All rights reserved.

2484

ative treatment of a variety of solid tumors,5 includinglocally advanced breast cancers,8-t 2 led the NSABP toinitiate a trial (B-18) to evaluate the worth of preoperativetherapy for primary operable breast cancer. The trial wasdesigned to determine whether such therapy more effec-tively prolonged disease-free survival (DFS) and survival(S) than the same chemotherapy given postoperatively;whether the response (regression) of a primary tumor orof palpable axillary lymph nodes to preoperative therapycorrelated with DFS and S outcomes; and whether suchtherapy, by reducing the size of a primary tumor, permit-ted more conservative breast surgery and decreased theincidence of ipsilateral breast tumor recurrence. Anotherobjective was to determine whether a downstaging ofaxillary lymph node status, ie, an increase in the propor-tion of node-negative patients, occurred.

In the B-18 study, 1,523 women were randomized totwo groups. Women in the first group underwent surgeryfollowed by adjuvant chemotherapy; those in the secondgroup received the same systemic therapy before surgery.This report focuses on the findings obtained from the 747eligible patients who received preoperative therapy anddocuments primary tumor response to adjuvant chemo-therapy administered before surgery. It also estimates theextent to which the performance of lumpectomy was in-creased when such therapy was administered. Informationis not yet available to compare the worth of preoperativeand postoperative therapy relative to DFS and S or todetermine whether a relationship exists between thoseoutcome parameters and primary tumor response to pre-operative therapy.

PATIENTS AND METHODS

Patient Eligibility

Women at participating NSABP institutions in the United Statesand Canada who had primary palpable, operable breast cancer(TI,2,3,N0,1,MO) were eligible for this study if their tumors wereconfined to the breast and axilla and if their life expectancy was atleast 10 years. Tumors had to be movable in relation to the underly-ing muscle and chest wall, and palpable axillary nodes could not befixed to each other or to the neurovascular bundle. Patients whosetumors demonstrated ulceration, erythema (inflammatory cancer),skin fixation, peau d'orange, satellite breast nodules, or parasternalnodules were ineligible.

Study Information and Characteristics of EligiblePatients at Randomization

Women were enrolled onto the study between October 1988 andApril 1993. Patients with breast cancer diagnosed by fine-needleaspiration or tru-cut biopsy (but not by incisional or excisional bi-opsy) who gave written consent to enter the study were stratifiedaccording to age (:s 49 years or > 50 years), clinical tumor size(< 2.0 cm, 2.1 to 5.0 cm, or > 5.1 cm), and clinical nodal status(negative or positive). Women were randomized to two treatment

FISHER ET AL

groups: the postoperative group, in which surgery (lumpectomy andaxillary node dissection, or modified radical mastectomy) was fol-lowed by systemic chemotherapy; or the preoperative group, inwhich the same systemic therapy was administered before surgeryand axillary node dissection. All women treated by lumpectomyalso received postoperative radiation therapy. After the cytologicdiagnosis of cancer and before randomization, investigators wererequired to indicate the type of operation (lumpectomy or mastec-tomy) they intended to perform. Therapy was administered postoper-atively to 763 of 1,523 patients randomized to this study; 760 patientsreceived preoperative therapy (Table 1). Four patients (0.5%) in thepostoperative group were ineligible: one had advanced disease; inanother, the time from biopsy to mastectomy exceeded protocollimits; the remaining two were ineligible for different reasons. Allbut 13 (98%) of the patients in the preoperative group were eligiblefor the study. Five of the 13 ineligible patients had advanced diseaseat randomization and three had an open biopsy; each one of theremaining five patients was ineligible for a different reason. In 1991,the NSABP Biostatistical Center discovered alterations in data thathad been submitted by an investigator from St Luc Hospital, Mon-treal, Canada. These alterations had been made before initial random-ization and were related to patient eligibility criteria. A total of 210of the 1,523 randomized patients were from St Luc; three had datairregularities. All three were declared ineligible and were managedin a manner similar to that of ineligible patients from other institu-tions. When all St Luc patients were omitted from the current analy-ses, no substantive differences in either results or conclusions wereobtained. The distribution of selected patient and tumor characteris-tics was similar in the two treatment groups (Table 1). Approxi-mately 50% of patients were less than 50 years of age and approxi-mately three fourths were clinically node-negative, ie, had nopalpable axillary nodes. Approximately 28% of the women in bothgroups had tumors - 2 cm in greatest diameter, and 13% had tumors> 5.1 cm.

Tumor Size Determination and Evaluation of

Preoperative Therapy Response

Clinical size of primary breast cancers and axillary nodes, if thelatter were palpable and considered to be suspicious for cancer, wasdetermined separately immediately before administration of each of

Table 1. Characteristics of Eligible Patients

Treatment GroupEligibility and Characteristics at

Randomization Postoperative Preoperative

Eligibility (no.)Randomized 763 760Eligible 759 747

Characteristics (%)Age, years

- 49 52 51- 50 48 49

Nodal status (clinical)Negative 74 74Positive 26 26

Breast cancer (clinical size, cm)- 2.0 27 292.1-5.0 59 58> 5.1 13 13Not reported < 1 0


PREOPERATIVE THERAPY FOR OPERABLE BREAST CANCER

four cycles (courses) of chemotherapy and before surgery. At eachassessment, the product of the two greatest perpendicular diametersof the tumors in the breast and axilla was calculated. The size ofbreast and axillary tumors was individually recorded for patientswho had palpable tumors in their axilla. The sum of both measure-ments was also recorded. The value of the combined measurementsrepresented the total or overall clinical tumor burden.

To have been considered assessable for tumor response, a patientmust have received at least two cycles of preoperative therapy, hadbidimensional measurements recorded at the beginning of cycle 1,and had at least one additional set of bidimensional measurementsrecorded at, or after, the end of cycle 2. More than 90% of assessablepatients received all four cycles of preoperative therapy and had allof the bidimensional measurements that were required. Approxi-mately two thirds of the women whose tumors could not be evaluatedwere classified as nonassessable because tumor response to therapyhad not been monitored according to protocol specifications. Tumorsof the remaining one third could not be evaluated because of devia-tions in protocol treatment.

After preoperative chemotherapy was administered, breast tumorswere classified according to clinical response. In the absence ofclinical evidence of tumor in the breast, the response to therapy wascategorized as clinically complete (cCR). When the reduction inclinical size of the breast tumor was - 50%, the response was judgedto be partial (cPR). When there was an increase of greater than 50%in the size of a breast tumor after a minimum of two cycles oftherapy, the patient was considered to have progressive disease(cPD). Patients whose breast tumor response did not meet the defini-tion of either cCR, cPR, or cPD were considered to have stabledisease (cSD). Thus, patients with cSD could have a tumor responseof less than 50% or an increase in tumor size of - 50%. A similarclassification was used to record the response of an axillary tumorto preoperative therapy in patients who had clinically positive nodeswhen they received their first cycle of preoperative therapy. Theoverall clinical response of both breast and axillary tumors wasdetermined by combining the sum of the product of the tumor mea-surements in both the breast and axilla. The development of a clini-cally suspicious ipsilateral axillary tumor during chemotherapy wasconsidered evidence of cPD in patients whose axilla was clinicallynegative when cycle 1 of the therapy was administered.

Surgical specimens from patients who had either a clinically com-plete breast or axillary tumor response to preoperative therapy wereevaluated for their pathologic tumor status. A pathologically com-plete response (pCR) was considered to have occurred when therewas no histologic evidence of tumor cells. The pathologic findingspresented in this report are those reported by institutional patholo-gists.

Therapy

All patients in both groups received doxorubicin and cyclophos-phamide (AC) therapy every 21 days at 60 and 600 mg/m2, respec-tively, for four courses. No dose reduction of either drug was permit-ted except following febrile neutropenia, where a 25% reduction inboth drugs was required. Administration of AC was delayed as aresult of either hematologic or gastrointestinal toxicity on day 1 ofany cycle. When permissible, drug administration was resumed atfull dose. All patients _ 50 years of age received tamoxifen, 10 mgtwice a day, for 5 years, beginning on the day after the last dose ofchemotherapy. Patients in the preoperative group received all fourcourses of chemotherapy, unless cPD occurred before completionof therapy. The remaining courses were administered after surgeryin women who developed cPD during chemotherapy, if these tumors

2485

were operable, and if the surgery had been performed before all fourcourses of chemotherapy were administered. Women whose tumorsbecame inoperable during therapy (cPD) were listed as treatmentfailures and were managed at the discretion of their physicians.

Radiation therapy was begun within 4 weeks after lumpectomyin women who received preoperative chemotherapy; women whoreceived postoperative chemotherapy underwent breast irradiationwithin 4 weeks after completion of their last course of therapy.Details of lumpectomy, axillary dissection, and radiation therapyhave been previously reported." All P values presented were basedon two-sided alternatives.

RESULTS

Tumor Response to Preoperative Therapy

Overall clinical local-regional tumor response. Of747 eligible patients assigned to preoperative therapy, 680(91%) were assessable for overall clinical local-regionaltumor response and 693 (93%) for tumor response in thebreast. The response of the axillary tumor to chemother-apy was assessable in 187 (93%) of 201 patients whosepalpable axillary nodes were considered suspicious forcancer. Response in the breast alone was assessable in194 (97%) patients, and overall response in the breastand axilla was assessable in 185 (92%). A cCR of bothbreast and axilla was documented in 238 patients (35%)and a cPR was documented in 302 (44%), for a total of540 (79%) patients who had a clinical response of bothbreast and axilla. An additional 114 (17%) patients hadcSD and 26 (4%) had cPD.

Clinical and pathologic response of primary breasttumors. Primary breast tumors from 693 patients wereevaluated for their response to preoperative therapy (Ta-ble 2). A reduction in tumor size occurred in 80% ofpatients: 36% had a cCR and 44% a cPR. Clinical tumorsize remained unchanged, decreased less than 50%, orincreased 50% or less in 17% of patients. Progression oftumor growth during preoperative therapy occurred inonly 3% of patients. The 23 patients with cPD in thebreast had tumors that were clinically operable; 12 pa-tients had a lumpectomy and 11 had a mastectomy. Onepatient who had a cPR was diagnosed with inoperableinflammatory cancer subsequent to a fourth cycle of ther-apy. Patient age failed to predict the clinical response ofbreast tumors (P = .15). The slightly greater number ofcomplete responses in younger patients was not statisti-cally significant (P = .18). Although the size of a breasttumor observed at the start of preoperative therapy wasunrelated to whether a tumor response occurred (P = .33),initial size was a predictor of complete tumor response (P< .001). Fifty-seven percent of patients with breast tu-mors :< 2.0 cm, 35% of those with tumors 2.1 to 5.0 cm,and 17% of those with tumors > 5.1 cm demonstrated acCR. Progressive disease occurred slightly more often inpatients who had smaller tumors. This was, perhaps, due


Table 2. Clinical Response of Breast Tumors to Preoperative Therapy

All Patientst Age, Years (%) Clinical Tumor Size,* cm (%) Clinical Nodal Status*t (%)

(n = 693) - 49 - 50 • 2.0 2.1-5.0 5.1 Negative PositiveClinical Response %) (n =353) (n = 340) (n = 131) (n = 445) (n 117) (n = 496) (n = 194)

Responders 80 82 78 79 81 75 78 85Complete 36 39 34 57 35 17 35 41Partial 44 43 44 22 46 58 43 44

Nonresponders 20 18 22 21 19 25 22 15Stable 17 15 18 15 16 22 18 13Progressive 3 3 4 5 3 3 4 2

P

Response (completeor partial) .15 .33 .04

Complete response .18 < .001 .16

"At cycle 1.

"tNot reported for 3 patients with assessable clinical tumor response in the breast.tRegardless of age, tumor size, or nodal status.

to the fact that, in those women, a slight increase in tumorsize could result in an increase of 50% or more in theproduct of the two greatest tumor diameters whereas, inwomen with larger tumors, a much greater increase insize was needed for the same increase to be observed.More patients who had clinically positive nodes displayedsome response to preoperative therapy (85% v 78%, re-spectively; P = .04). The number of breast tumors thatcompletely responded to preoperative therapy was greaterin women with clinically positive nodes than in womenwith negative nodes, but the difference was not significant(41% v 35%, P = .16).

When evaluated by multivariate modeling using logis-tic regression, predictors of clinical response were similarto those observed in univariate analyses. When age, nodalstatus, and tumor size were entered into the model, onlynodal status remained a significant (P = .02) predictor ofeither complete or partial tumor response to preoperativetherapy. Node-positive patients were more likely to havea tumor response. There was a suggestion of improvedresponse for smaller tumors and for younger patients, butneither characteristic was a significant predictor. WhencCR was the outcome variable, tumor size (P < .0001)and clinical nodal status (P = .003) were highly signifi-cant independent prognostic variables, whereas age wasof borderline significance (P = .08). Younger patientswith clinically positive nodes and small tumors were mostlikely to have a cCR.

Having observed that clinical tumor size and clinicalnodal status are independent prognostic variables withregard to cCR, the rate of cCR in each of the six patienttumor-size/nodal-status subsets is of interest (Table 3).However, it should be noted that the number of patientsvaries across subset and is particularly small (n = 17) in

the group of node-positive patients with tumor size lessthan 2.0 cm. The cCR rate of smaller tumors was greaterthan that of larger tumors, regardless of axillary nodalstatus. However, the likelihood of having a cCR wasgreater when smaller tumors (s 2.0 cm) were associatedwith positive nodes (82% v 54%). These findings weresimilar when tumors were 2.1 to 5.0 cm in size. In node-positive patients, the cCR was 44%; in node-negativepatients, it was 32%. However, the cCR rates were nearlyequal in women with the largest tumors (- 5 cm), regard-less of whether their nodes were positive or negative (16%v 19%, respectively).

Pathologic examination of breast tissue from 245women whose tumors demonstrated a cCR to preopera-tive therapy showed no evidence of cancer (a pCR) in26% and only noninvasive tumor (ductal carcinoma insitu [DCIS]) in 11% (Table 4). When the incidence ofpCR pathologic findings was related to the clinical sizeof tumors present in this group of patients when the firstcycle of chemotherapy was begun, the pCR of tumorswas no greater in women who initially had small tumorsthan in those who had large tumors. When the frequency

Table 3. Complete Clinical Response of Breast Tumors to Preoperative

Therapy According to Tumor Size and Nodal Status

Clinical Tumor Size, cm

Clinical Nodal Status < 2.0 (cCR) 2.1-5.0 (cCR) I 5 (cCR)

Negative% 54 32 19

No. 114 312 70

Positive

"% 82 44 16

No. 17 132 45

2486 FISHER ET AL



Table 4. Pathologic Status of Breast Tumors Demonstrating Complete Clinical Response to Preoperative Therapy According to Clinical Tumor Size

and Clinical Nodal Status

Complete Clinical Response of Breast Cancer (%)

Clinical Tumor Size, cm* Clinical Nodal Status*

All Patients • 2.0 2.1-5.0 • 5.1 Negative PositivePathologic Status of Operative Speciment (n = 245) (n = 73) (n = 153) (n = 19) (n = 168) (n = 77)

Free of tumor 26 25 27 21 24 30

Noninvasive tumor (DCIS) present 11 4 14 5 8 16

Invasive tumor present 64 71 59 74 68 55

Abbreviation: DCIS, ductal carcinoma in situ.

*At onset of preoperative therapy (cycle 1).

tExcludes 7 unknowns.

of pCR was related to initial nodal status of breasttumors in women who had a cCR, a slightly greaterpercentage of tumors in women with breast cancersassociated with clinically positive nodes demonstrateda pCR (30%) than did tumors in women with clinicallynegative nodes (24%; P = .35). More resected speci-mens from women with clinically positive nodes werefound on pathologic examination to contain noninva-sive breast tumors (16%) than did specimens fromwomen with clinically negative nodes (8%; P = .12).Of particular significance was the finding that, in breasttumors associated with clinically positive nodes, almostone half (45%) of the women whose tumors demon-strated a cCR to preoperative therapy had no invasivecancer on pathologic examination, as compared withalmost one third (32%) of clinically node-negative pa-tients who had a cCR (P = .05).

The evaluation relative to pCR in the breast was limitedto women who attained cCR on the assumption that pCRwould not occur in women with cPR, cSD, or cPD. Basedon this assumption, pCR in the breast occurred in only9% of patients treated with preoperative therapy. (Anadditional 4% were found to have only noninvasive tu-mor, which resulted in a total of 13% who were free ofinvasive tumor.) Fourteen percent of patients with pCR

had tumors - 2.0 cm, 9.3% had tumors 2.1 to 5.0 cm,and 3.4% had tumors - 5.1 cm.

Clinical and pathologic response of clinically positiveaxillary nodes. Eighty-nine percent of patients withclinically positive axillary nodes had a clinical nodal re-sponse to preoperative therapy (Table 5). In 73%, theclinical response was complete, and in only 2% did cPDoccur during therapy. Pathologic examination of speci-mens obtained after axillary dissection in patients whohad a cCR showed a greater rate of pCR and a greaterdecrease in the number of positive nodes than was evidentin patients who had a cPR.

When the 185 patients with clinically positive axillarynodes were evaluated relative to the overall clinical responseof their tumor, ie, with regard to the sum of the product ofthe measurement of their breast and axillary tumors at theonset of preoperative therapy (cycle 1), 85% demonstrateda response to preoperative therapy. A cCR of both breastand axillary nodes was observed in 66 (36%) of 185 patientswith clinically positive nodes. Twenty percent of the 66patients who had a cCR in both the breast and axilla, or7% of the 185 with clinically positive nodes, were foundon pathologic examination to have a pCR in both the re-sected breast and axillary tissue. Progressive disease wasnoted in only three (2%) of 185 patients.

Table 5. Clinical and Pathologic Response of Clinically Positive Axillary Nodes to Preoperative Therapy

Pathologic Response (% of patients)

Type of Clinical No. of Patients No. of Positive Nodes in Axillary Specimen

Response (n = 185)* Negative Positive 1-3 4-9 10

Responders 164 40 60 31 21 8Complete 135 44 56 32 17 7Partial 29 17 83 28 41 14

Nonresponders 21 24 76 29 24 24Stable 18 22 78 28 28 22Progressive 3 t t t t t

*Excludes 2 patients with unknown pathologic response.tDenominator too small to obtain meaningful information.

2487


Table 6. Comparison Between Postoperative and Preoperative Therapy Groups Relative to Clinical Nodal Status, Clinical Tumor Size,and Pathologic Nodal Status

Regardless of Clinical Clinical Nodal Status* (%) Clinical Tumor Size,* cm (%)Nodol Status or Tumor

Size (%) Negative Positive 2 - 2.1

Postop Preop Postop Preop Postop Preop Postop Preop Postop PreopPathologic Nodal Statust (n = 743) (n = 735) (n = 554) (n = 543) (n = 189) (n = 192) (n = 200) (n = 211) (n = 540) (n = 524)

Negative 43 59 52 67 14 36 54 73 39 54

Positive 57 41 48 33 86 64 46 27 61 46

1-3 30 24 29 21 34 33 30 18 30 27

4-9 17 12 14 9 28 21 12 7 19 14

- 10 10 4 5 2 23 9 4 1 12 5

P value, pathologic

nodal status

(negative/positive) < .001 < .001 < .001 .001 < .001

Abbreviations: Postop, postoperative; Preop, preoperative.

'Before randomization (excludes 3 patients with unknown tumor size in the postoperative group)."tExcludes 16 patients in the postoperative group and 12 patients in the preoperative group with unknown status.

Comparison of the postoperative and preoperative groupsrelative to clinical nodal status or to clinical tumor sizeand pathologic nodal status. When each of the treatmentgroups was compared relative to their pathologic nodal sta-tus regardless of their clinical nodal status or tumor size,43% of patients in the postoperative group were found tobe pathologically node-negative, whereas 59% of patientsin the preoperative group had no pathologic evidence oftumor in their nodes (P < .001; Table 6). Fewer positivenodes were found in patients who received preoperativetherapy.

When clinical nodal status was compared with patho-logic nodal status, only approximately one half (52%) ofpatients in the postoperative group who had been consid-ered to be clinically node-negative were found to bepathologically node-negative. In contrast, in the preopera-tive group, two thirds (67%) of patients judged beforerandomization to be clinically node-negative were foundto be pathologically node-negative (P < .001). Thirty-six percent of patients initially considered to be node-positive were found to be pathologically node-negativeafter preoperative therapy, compared with 14% of thosein the postoperative therapy group (P < .001). The num-ber of positive nodes in clinically node-negative patientswho received preoperative therapy was less than that inclinically node-negative patients who received postopera-tive therapy. Similarly, the percent of clinically node-positive patients who received preoperative therapy andwho were found to have four to nine or > 10 positivenodes was less than that observed in the group treatedwith postoperative therapy. The proportion of patientswith one to three positive nodes was similar in bothgroups.

When patients were evaluated according to clinical tu-

mor size and treatment group, it was found that a greaterproportion of patients in the preoperative group hadpathologically negative nodes, regardless of whether theirtumors were : 2 cm (P = .001) or - 2.1 cm (P < .001)in size. Seventy-three percent of patients with tumors of-! 2 cm in the preoperative group were pathologicallynode-negative, whereas 54% of similar patients in thepostoperative group had pathologically negative nodes.In patients with tumors - 2.1 cm, 54% in the preoperativegroup were pathologically node-negative, as comparedwith 39% in the postoperative group. There were alsofewer patients in each positive-node category after preop-erative therapy; this was particularly evident in womenwhose tumors were -z 2 cm.

Comparison Between the Postoperative andPreoperative Groups Relative to Frequency ofLumpectomy Proposed and Peiformed

Proposed. The type of operation proposed (ie, lump-ectomy or mastectomy) was specified for each patientbefore randomization. Lumpectomy was proposed in ap-proximately two thirds of all eligible patients in bothtreatment groups (Table 7). It was the treatment specifiedfor 86% of all patients with tumors :!s 2 cm, ie, 83% in thepostoperative group and 89% in the preoperative group.Lumpectomy was specified for 70% of patients with tu-mors of 2.1 to 5.0 cm in the postoperative and preopera-tive groups (71% and 68%, respectively). It was the pro-posed procedure for only a few women (3%) with tumors>- 5.1 cm. More clinically node-negative than clinicallynode-positive patients were considered candidates forlumpectomy. That situation also prevailed in women withtumors - 2.0 cm or 2.1 to 5.0 cm. Multivariate modelingindicated that clinical tumor size was highly significant

2488 FISHER ET AL



Table 7. Comparison Between Postoperative and PreoperativeTherapy Groups Relative to Frequency of Lumpectomy Proposed and

Performed According to Nodal Status and Tumor Size

Lumpectomy LumpectomyClinical Nodal Status Proposed* (%) Performed (%)

and Tumor Size(cm)* Postop Preop Postop Preop

All patients 66 65 60 67- 2.0 83 89 79 81

2.1-5.0 71 68 63 71S5.1 3 3 8 22

Negative 70 71 64 70< 2.0 85 90 80 832.1-5.0 73 73 65 73

S5.1 4 4 8 14Positive 53 49 48 59

i 2.0 74 82 76 682.1-5.0 67 57 56 66- 5.1 2 2 9 33

*At time of patient entry and before randomization.

(P < .0001) for predicting when a lumpectomy wouldbe proposed. As tumor size increased, the likelihood oflumpectomy being proposed decreased; lumpectomy wasrarely proposed in women with tumors greater than 5 cm.Nodal status was also prognostic (P = .02) in that node-negative patients were more often considered for lumpec-tomy than were patients with positive nodes. Age wasnot a factor.

Performed. When the two groups were comparedwith respect to type of operation performed, the frequencyof lumpectomy was greater in the preoperative group(60% v 67%; P = .002) (Table 7). This was particularlyevident in patients with tumors -- 5.1 cm (8% in thepostoperative group v 22% in the preoperative group).Lumpectomy was performed more often in women withclinically negative nodes than in women with clinicallypositive nodes (64% v 48%, respectively, in the postoper-ative group, and 70% v 59%, respectively, in the preopera-tive group). In both groups, clinical tumor size was highlyprognostic for the type of surgery performed (P < .0001).After adjustments were made for clinical tumor size inthe preoperative group, it was found that patients with acCR were more likely than those with a cPR to have hada lumpectomy (P = .01). Women with a cPR were morelikely to have had a lumpectomy than were women whodid not respond to therapy (P = .01).

When the overall frequency of lumpectomy proposedwas compared with that performed (Table 7), 6% fewerpatients in the postoperative group had a lumpectomythan was originally proposed (60% v 66%) whereas, inthe preoperative group, the percent of patients who under-went a lumpectomy was about the same as the percentof patients for whom lumpectomy was proposed (67%v 65%, respectively). When the frequency with which

2489

lumpectomy was performed was considered relative totumor size, particularly in patients with tumors - 5.1 cm,there was clear evidence that more lumpectomies wereperformed (22%) than proposed (3%). When examinedaccording to clinical nodal status, the incidence of lump-ectomy performed after preoperative therapy was greaterthan that proposed in node-positive patients (59% v 49%,respectively); this was particularly evident in womenwhose tumors were - 5.1 cm in size (33% v 2%, respec-tively).

DISCUSSION

Although the neologism "neoadjuvant or anterior"was introduced in 1981 to depict the novel strategy ofusing preoperative chemotherapy to treat head and neckcancers and osteosarcoma,' 4 and although studies re-porting the use of such therapy for the management oflocally advanced breast cancer appeared in the early1980s,812 when NSABP B-18 was implemented in 1988,only three publications considered the use of preoperativesystemic therapy in patients with primary operable tu-mors. Jacquillat et al"t •2 first reported the use of a multi-drug regimen as primary therapy in women with tumorsranging from clinical stage I to stage IIIB. Shortly thereaf-ter, Forrest et al'5 reported findings obtained from 29women with large operable breast cancers who receivedendocrine therapy or chemotherapy as their initial treat-ment, and Ragaz'6 described the design of a trial to evalu-ate preoperative therapy.

Because of the paucity of information obtained fromthose studies, the NSABP initiated a randomized clinicaltrial (B-18) to evaluate the validity of a hypothesis that, ifconfirmed, would lend support to the use of preoperativetherapy for the treatment of operable breast cancer. De-spite the popularity of that concept, little information hasbeen reported since B-18 was begun. A majority of thefindings are from studies conducted in heterogeneousgroups of nonrandomized patients.7-24 Only four reportspublished since 1988 relate to randomized trials con-ducted to obtain information regarding the worth of pre-operative therapy. 2 -28 Unfortunately, the manner in whichthese studies were designed prohibits a straightforwardcomparison between patients who received only preoper-ative chemotherapy and those who received the sametherapy postoperatively. Although these studies providelimited information to indicate that a proportion of pri-mary breast cancers clinically regress after the administra-tion of diverse chemotherapeutic and/or hormonal agents,they fail to provide the kind of information that we presentin this report.

The findings from B- 18 have been obtained from morepatients (n = 693) than have been enrolled in all of the


FISHER ET AL

previously reported randomized trials combined, frommore women than were in six of the seven nonrandom-ized studies, and from more than the combined numberof patients in the two major nonrandomized studies fromMilan.21 '22 Our findings that 80% of patients demon-strated a decrease in the size of their breast tumors arein keeping with those reported from the disparate studiesconducted by other investigators, who noted a primarytumor response to preoperative therapy that varied from63% to 85% (median, 74%).22 The percent of patients inour study who had a cCR (36%) was higher than thatnoted in other studies (range, 12% to 30%; median,19%).22

The findings in this report relating age, clinical tumorsize, and clinical nodal status to clinical response of breasttumors treated by preoperative therapy are of particularsignificance because other studies have provided littleinformation in that regard.15,18,2 '1,23 Our findings indicatethat, when evaluated regardless of the initial size of theirbreast cancer, approximately 80% of patients demon-strated a decrease of > 50% in their tumor size. However,there was a strong correlation between tumor size andthe incidence of a cCR. Because more women withsmaller tumors had a cCR to preoperative therapy, itmight have been expected that a cCR would also occurmore often in women with negative axillary nodes. Con-versely, we observed a cCR more frequently in node-positive than in node-negative patients. Of particular in-terest were our observations that, when examined relativeto tumor size and nodal status, small tumors associatedwith positive nodes had a greater percentage of cCR thandid small tumors associated with negative nodes. On theother hand, the frequency of cCR was similar in womenin both nodal groups who had the largest breast tumors,ie, those > 5 cm.

Only one trial that evaluated preoperative therapy useda multivariate analysis to determine independent prognos-tic factors for DFS or S outcome.' 8 However, none of thestudies have assessed independent factors responsible fora primary breast cancer clinical response to preoperativetherapy. In the B-18 trial, multivariate analysis indicatedthat clinical tumor size (smaller tumors) and clinical nodalstatus (positive nodes) were independent predictors ofcCR in the breast, whereas age (younger patients) was ofborderline significance. Only clinical nodal status wasfound to be a significant prognostic factor for breast tumorresponse.

There is general agreement that knowledge regardingthe clinical response to preoperative therapy of a primarybreast tumor and/or clinically involved axillary nodescould be important, particularly if such therapy leads tomore breast-conserving surgery. Should a breast cancer

response to preoperative therapy mirror the effect of suchtherapy on concomitant occult metastases, that knowledgeis apt to be of even greater significance. The response ofa primary breast tumor to preoperative therapy could thenbe used as a marker to predict the response of distantmicrometastatic disease to systemic therapy, thus elimi-nating the need to wait many years for information onpatient outcome before determining the efficacy of a par-ticular therapy. All current knowledge about the responseof metastases to systemic therapy has been obtained fromobservation of the response of overt metastatic disease.Most often, as is observed in advanced disease, the re-sponse of metastases to therapy is only partial and thebenefits short-lived. Even when a response assessed clini-cally, radiologically, or by a particular biologic markeris considered to be complete, regrowth of metastases oftenoccurs, an indication of the failure of the therapy to elimi-nate all tumor cells. Similarly, the occurrence of a cCRwith an incomplete pathologic response subsequent to theuse of preoperative therapy may be associated with onlya partial eradication of occult metastases. Perhaps distantdisease can be eradicated only by a preoperative regimenthat completely eliminates a primary breast tumor patho-logically. Should that be the case, ie, should a pCR topreoperative therapy prove to be of greater importancethan is a cCR, then our finding of the infrequency of apCR emphasizes the need for better therapies to increasethe incidence of a pCR. In that regard, more prolongedadministration of AC therapy and/or the use of otherregimens may substantially increase the number of pa-tients with a pCR. To test this thesis, the NSABP iscurrently conducting a study (B-27) to evaluate the ad-ministration of Taxotere (Rh6ne-Poulenc Rorer, Col-legeville, PA) after completion of the same AC regimenas was used in B-18.

A major aim of the B-18 study was to determine if,by decreasing the size of a primary tumor, preoperativechemotherapy permitted the performance of morelumpectomies. Although a majority of primary tumorsdecrease in size after preoperative therapy, whether thatfinding has, indeed, resulted in a major increase inbreast-conserving operations, as other investigatorshave noted, remains speculative. When evaluating re-

ports that purport to demonstrate that such therapy in-creases the use of lumpectomy, one must keep in mindthat the frequency with which lumpectomy is per-formed may vary widely because of differences in thecriteria used to decide which patients are suitable forsuch treatment.' A large proportion of patients consid-ered by other investigators 21-23 to have been spared theneed for breast removal because they had received pre-operative therapy would, by our standards, have been

2490



eligible for lumpectomy even if they had not receivedsuch therapy. To provide some idea of the extent towhich preoperative therapy increased the likelihoodthat lumpectomy would be performed in the B-18study, surgeons were asked to indicate, before random-ization, the type of operation they proposed.

Before randomization, lumpectomy was specified asthe proposed procedure in two-thirds of the 1,506 pa-tients eligible in the B-18 study. Univariate and multi-variate analyses indicated that both tumor size andnodal status influenced a surgeon's decision. There wasa higher likelihood of lumpectomy being proposed inclinically node-negative patients. That selection pre-vailed within any tumor-size category, an indicationthat many physicians considered that lumpectomy wasinappropriate for node-positive patients, even whentheir tumors were < 2 cm in size. Age was not foundto be a factor in the decision to propose lumpectomy.This finding was in contrast to evidence in the literatureindicating that breast conservation is less often used inolder women because of physician bias.29

The best evidence to indicate that preoperative ther-apy results in an increase in the frequency with whichlumpectomy is performed comes from our finding of asevenfold increase in the performance of lumpectomyover that proposed in patients with tumors > 5.1 cm.Whereas lumpectomy was proposed in only 2% of thewomen with tumors > 5 cm who were judged beforepreoperative therapy to be clinically node-positive, itwas performed in 33% of patients after preoperativetherapy and in only 9% after postoperative therapy.Moreover, there was also an increase in breast-con-serving surgery after preoperative therapy in node-negative patients whose tumors were - 5.1 cm in size(14%, compared with 8% in the postoperative therapygroup). It may be cogently asked why, in B-18, mas-tectomy was proposed in 14% of patients with tumors< 2.0 cm and in 30% of patients with tumors 2.1 to5.0 cm, or why lumpectomy was not performed in21% and 37% of postoperative patients who had theformer and latter tumor sizes, respectively. Aside fromphysician bias against lumpectomy, the failure to per-form that operation may have been due to some sur-geons' belief that the breast size of many of the womenwas too small to permit a favorable cosmetic result.

Another objective of the B-18 study was to provideinformation regarding whether preoperative therapyresulted in a downstaging of axillary nodal status andan increase in node-negative patients. Such a findingcould provide support for the elimination of axillarynodal dissection, at least in certain patient cohorts.

2491

Our current findings continue to indicate, as we re-ported in 1981,30 that there is significant inaccuracyin the clinical staging of axillary nodes. These resultsserve as a reminder that clinical nodal status is a poormeasure of pathologic nodal status; thus, analysis ofclinical nodal status must be interpreted with caution.The B-18 findings indicating that 48% of women withclinically negative nodes were pathologically node-positive and that 14% of women with clinically posi-tive nodes were pathologically node-negative may beused by those who favor axillary dissection as evi-dence for its propriety. That preoperative therapy sig-nificantly increased the incidence of pathologicallynegative axillary nodes is apparent from a comparisonof the nodal status of women in the two treatmentgroups. When patients in both groups were comparedrelative to the pathologic status of their axillary nodes,there was a 37% increase in the incidence of patholog-ically negative nodes, ie, a 29% increase in patientswhose nodes were considered to be clinically nega-tive, and a 157% increase in patients considered tohave clinically positive nodes. The number of patientsin the preoperative group who had one to three, fourto nine, or t 10 pathologically positive nodes wasless than that in the postoperative therapy group.

Because we have shown that clinical nodal status,clinical tumor size, and treatment are independent vari-ables for predicting the presence of pathologically neg-ative nodes, it might be considered that axillary dissec-tion could be eliminated in patients who are clinicallynode-negative, who have small tumors, and who re-ceive preoperative therapy. However, our findings insuch patients indicate that 26% of clinically node-neg-ative patients with tumors of < 2.0 cm had pathologi-cally positive nodes after preoperative therapy. Al-though this was less than the number of node-positivepatients in the postoperative therapy group (40%), thatincidence may be viewed by those who continue tofavor the use of axillary dissection as being too highto permit avoiding axillary dissection. Whether the useof tamoxifen further downstages nodal status cannotbe ascertained from B-18 because tamoxifen was ad-ministered to patients aged - 50 years after comple-tion of preoperative therapy.

When all of the findings presented in this report aretaken into consideration, it remains our contention'that, until the findings from B-18, or from some otherrandomized trial, demonstrate that preoperative therapyresults in distant DFS and S equivalent or superior tothat observed after postoperative therapy, administer-ing preoperative therapy in women amenable to treat-


2492 FISHER ET AL

ment by lumpectomy without requiring the use of such preoperative therapy in an attempt to eliminate the need

therapy to reduce the size of their tumors seems inap- for breast removal.

propriate. When there is concern that a satisfactory ACKNOWLEDGMENTcosmetic result and/or control of the disease in the We thank Tanya Spewock for editorial assistance, Mary Hof forbreast might not be achieved by lumpectomy, we rec- preparation of the manuscript, and Fei-Na Chen for assistance withommend that women be given an opportunity to receive the analysis.

APPENDIXInstitutions Contributing 10 or More Patients to NSABP B-18

Institution Principal Investigator

Boston Medical Center, Boston, MA Maureen T. Kavanah, MDCCOP, Alton Ochsner Medical Foundation, New Orleans, LA Carl G. Kardinal, MDCCOP, Columbia River Oncology Program, Portland, OR Keith S. Lanier, MDCCOP, Kaiser Permanente, San Diego, CA Scott M. Browning, MDCCOP, Marshfield Clinic, Marshfield, WI James L. Hoehn, MDCross Cancer Institute, Edmonton, Canada Alan W. Lees, MDEisenhower Medical Center, Rancho Mirage, CA Phillip D. Bretz, MDHarbor-UCLA Medical Center, Torrance, CA Michael J. Stamos, MDHopital Maisonneuve-Rosemont, Quebec, Canada Yves Eugene LeClerc, MDHotel-Dieu, Montreal, Canada Andre Robidoux, MDJewish General Hospital, Montreal, Canada Richard G. Margolese, MDKent County Memorial Hospital, Warwick, RI Peter Baute, MDL'Hopital Laval, Quebec, Canada Martin Potvin, MDMBCCOP, MCV/CMH of Virginia, Richmond, VA Christopher E. Desch, MDMedical College of Virginia, Richmond, VA (CGOP) Harry D. Bear, MDMedical College of Wisconsin, Milwaukee, WI Alonzo P. Walker, MDMichigan State University, East Lansing, MI Nikolay V. Dimitrov, MDMount Sinai Medical Center, Cleveland, OH Richard S. Bornstein MD and Lawrence Levy, MDNE Ontario Regional Cancer Center, Sudbury, Canada Stephen Gluck, MDRoyal Victoria Hospital, Montreal, Canada Henry Shibata, MDSt. Luc Hospital, Montreal, Canada Andre Robidoux, MDSt. Michael's Hospital, Toronto, Canada Jarley Koo, MDSt. Sacrement Hospital, Quebec City, Canada Luc Deschenes, MDTom Baker Cancer Centre, Calgary, Canada A.H.G. Paterson, MDTufts Univ., New England Medical Center, Boston, MA (CGOP) Daniel David Karp, MDTulane University, New Orleans, LA Carl M. Sutherland, MDUniversity of California, Davis, CA Frederick J. Meyers, MDUniversity of Cincinnati, OH Bernard S. Aron, MDUniversity of Mississippi, Jackson, MS Terrence J. Hall, MD., PhDUniversity of Pittsburgh, PA Bernard Fisher, MDUniversity of Texas, San Antonio, TX Anatolio B. Cruz, Jr, MDUniversity of Vermont, Burlington, VT (CGOP) David N. Krag, MD

NOTE. A listing of institutions that contributed fewer than 10 patients is available from NSABP Headquarters.Abbreviations: CCOP, Community Clinical Oncology Program; MBCCOP, Minority-Based Community Clinical Oncology Program;

GCOP, Cooperative Group Outreach Program.

REFERENCES

1. Fisher B, Mamounas EP: Preoperative chemotherapy: A model apy for squamous cell cancer (SCC) of the esophagus: A potentiallyfor studying the biology and therapy of primary breast cancer. J Clin curative approach. J Clin Oncol 2:75-79, 1984Oncol 13:537-540, 1995 6. Rosen G: Preoperative chemotherapy for osteogenic sarcoma.

2. Gunduz N, Fisher B, Saffer EA: Effect of surgical removal on the Cancer 49:1221-1230, 1982growth and kinetics of residual tumor. Cancer Res 39:3861-3865, 1979 7. Schuller DE: Preoperative reductive chemotherapy for locally

3. Fisher B, Gunduz N, Saffer EA: Influence of the interval be- advanced carcinoma of the oral cavity, oropharynx, and hypophar-tween primary tumour removal and chemotherapy on kinetics and ynx. Cancer 51:15-19, 1983growth of metastases. Cancer Res 43:1488-1492, 1983 8. Shick P, Goodstein J, Moor J, et al: Preoperative chemotherapy

4. Goldie JH, Coldman AJ: A mathematical model for relating followed by mastectomy for locally advanced breast cancer. J Surg

the drug sensitivity of tumors to their spontaneous mutation rate. Oncol 22:278-282, 1983Cancer Treat Rep 63:1727-1733, 1979 9. Sorace RA, Bagley CS, Lichter AS, et al: The management of

5. Leichman L, Seydel HG, Steiger Z: Preoperative adjuvant ther- nonmetastatic locally advanced breast cancer using primary induc-



tion chemotherapy with hormonal synchronization followed by radi-ation therapy with or without debulking surgery. World J Surg 9:775-785, 1985

10. Perloff M, Lesnick J: Chemotherapy before and after mastec-tomy in stage III breast cancer. Arch Surg 117:879-881, 1982

11. Jacquillat CI, Baillet F, Blondon J, et al: Preliminary resultsof neoadjuvant chemotherapy in initial management of breast cancer(BC). Proc Am Soc Clin Oncol 2:112, 1983 (abstr)

12. Jacquillat CI, Weil M, Auclerc G, et al: Neoadjuvant chemo-therapy in the conservative management of breast cancers: Study of143 patients, in Ragaz J, Band PR, Goldie JH (eds): Recent Resultsin Cancer Research. New York, NY, Springer-Verlag, 1986, pp 113-119

13. Fisher B, Wolmark N, Fisher ER, et al: Lumpectomy andaxillary dissection for breast cancer: surgical, pathological and radia-tion considerations. World J Surg 9:692-698, 1985

14. Frei E: Clinical cancer research: An embattled species. Cancer50:1979-1992, 1982

15. Forrest APM, Levack PA, Chetty U, et al: A human tumourmodel. Lancet 2:840-842, 1986

16. Ragaz J: Emerging modalities for adjuvant therapy of breastcancer: Neoadjuvant chemotherapy. NCI Monogr 1:145-152, 1986

17. Mansi JL, Smith IE, Walsh G, et al: Primary medical therapyfor operable breast cancer. Eur J Cancer Clin Oncol 25:1623-1627,1989

18. Jacquillat C, Weil M, Baillet F, et al: Results of neoadjuvantchemotherapy and radiation therapy in the breast-conserving treat-ment of 250 patients with all stages of infiltrative breast cancer.Cancer 66:119-129, 1990

19. Well M, Borel C, Auclerc G, et al: Nonsurgical approach instage I and stage II breast cancer. Cancer Invest 10:581-586, 1992

20. Forrest APM, Anderson EDC: Primary medical therapy: Ed-inburgh experience, in Powles TJ, Smith IE (eds): Medical Manage-

2493

ment of Breast Cancer. Philadelphia, PA, Lippincott, 1991, pp 273-280

21. Bonadonna G, Veronesi U, Brambilla C, et al: Primary che-motherapy to avoid mastectomy in tumors with diameters of threecentimeters or more. J Natl Cancer Inst 82:1539-1545, 1990

22. Bonadonna G, Valagussa P, Zucali R, et al: Primary chemo-therapy in surgically resectable breast cancer. CA 45:227-243, 1995

23. B61embaogo E, Feillel V, Chollet P, et al: Neoadjuvant che-motherapy in 126 operable breast cancers. Eur J Cancer 28A:896-900, 1992

24. Smith LE, Jones AL, O'Brien MER, et al: Primary medical(neo-adjuvant) chemotherapy for operable breast cancer. Eur J Can-cer 29A:1796-1799, 1993

25. Mauriac L, Durand M, Avril A, et al: Effects of primarychemotherapy in conservative treatment of breast cancer patientswith operable tumors larger than 3 cm: Results of a randomized trialin a single centre. Ann Oncol 2:347-354, 1991

26. Scholl SM, Fourquet A, Asselain B, et al: Neoadjuvant versusadjuvant chemotherapy in premenopausal patients with tumours con-sidered too large for breast conserving surgery: Preliminary resultsof a randomised trial: S6. Eur J Cancer 30A:645-652, 1994

27. Van de Velde CJH: Preoperative chemotherapy in operablebreast cancer: The influence of timing FEC in relation to surgery.Drugs 45:31-37, 1993 (suppl 2)

28. Powles TJ, Hickish TF, Makris A, et al: Randomized trial ofchemoendocrine therapy started before or after surgery for treatmentof primary breast cancer. J Clin Oncol 13:547-552, 1995

29. Fisher B, Ore L: On the underutilization of breast-conservingsurgery for the treatment of breast cancer. Ann Oncol 4:96-98, 1993

30. Fisher B, Wolmark N, Bauer M, et al: The accuracy of clinicalnodal staging and of limited axillary dissection as a determinant ofhistological nodal status in carcinoma of the breast. Surg GynecolObstet 152:765-772, 1981


91 day challenge (preview)

Documents