Relation of Number of Positive Axillary Nodes to the

Prognosis of Patients With Primary Breast Cancer

An NSABP Update

BERNARD FISHER, MD. MADELINE BAUER, PHD,t D. LAWRENCE WICKERHAM, MD,$

CAROL K. REDMOND, S C D , ~AND EDWIN R . FISHER, MD,II
WITH THE CONTRIBUTION OF
ANATOLIO B. CRUZ. MD,T ROGER FOSTER, MD,# BERNARD GARDNER. MD. HARVEY LERNER. MD.tf
RICHARD MARGOLESE. MD** ROGER POISSON, MD.55 HENRY SHIBATA, MD,lI 11 HERBERT VOLK, MD,VV
AND OTHER NSABP INVESTIGATORS##

The current findings completely affirm the validity of our original observations indicating the appro-
priateness of grouping primary breast cancer patients into those with negative, 1 to 3, or 2 4 positive
nodes. Results, however, reveal that there is a risk in combining all patients with 2 4 positive nodes
into a single group. Since there was a 25% greater disease-free survival and an 18%greater survival
in those with 4 to 6 than in those with 213 positive axillary nodes, such a unification may provide
misleading information regarding patient prognosis, as well as the worth of a therapeutic regimen when
compared with another from a putatively similar patient population. Of particular interest were findings
relating the conditional probability, i.e., the hazard rate, of a treatment failure or death each year
during the 5-year period following operation to nodal involvement with tumor. Whereas the hazard
rate for those with negative, or 1 to 3 positive nodes, was relatively low and constant, in those with
2 4 positive nodes the risk in the early years was much greater, but by the fifth year it was similar to
that occurring when 1-3 nodes were involved, and not much different from negative node patients. The
same pattern existed whether 4 to 6 or 213 nodes were positive. When the current findings are
considered relative to other factors with predictive import, it is concluded that nodal status still remains
the primary prognostic discriminant.
Cancer 52: 1551-1 557. 1983.

P RIOR to 1968, the prognosis of patients with primary

breast cancer was related to the presence or ab-
sence of axillary lymph node involvement, with no re-
gard for the number of nodes demonstrating tumor. In
that year, findings were reported from the first clinical
trial of the National Surgical Adjuvant Breast and Bowel
Project (NSABP) indicating that increasing numbers of

From the National Surgical Adjuvant Breast and Bowel Project

(NSABP) Headquarters Room 9 14 Scaife Hall, 3550 Terrace Street,
Pittsburgh, Pennsylvania 1526 1, where reprints should be addressed.
Supported by USPHS grants R-OI-CA- 12027, contract NO 1-CB-
23876, and American Cancer Society grant RC-13
* Professor of Surgery, Department of Surgery, School of Medicine,
University of Pittsburgh, Pittsburgh, PA.
Statistician, NSABP Biostatistical Center, University of Pitts-
burgh, Pittsburgh, PA.
$ NSABP Fellow, Department of Surgery, School of Medicine, Uni-
versity of Pittsburgh, Pittsburgh, PA.
0 Professor of Biostatistics, Graduate School of Public Health, Uni-
versity of Pittsburgh, Pittsburgh, PA.
I( Chief, Department of Pathology, Shadyside Hospital, Pittsburgh,
PA.
11 Professor of Surgery, Department of Surgery, University of Texas,
Medical School at San Antonio, San Antonio, TX.
# Professor ofsurgery, University of Vermont, College of Medicine,
Burlington, VT.
** Professor of Surgery, Director, Tumor Clinic, Downstate Med-
ical Center, State University of New York, Brooklyn, NY.
??Head, Section on Surgical Oncology and Cancer Chemotherapy.
Department of Surgery, Pennsylvania Hospital. Philadelphia. PA.
$$ Director, Division of Oncology, Department of Surgery, Jewish
General Hospital, Montreal, Quebec, Canada.
$9 Head of the Service of Surgical Oncology, Director of the On-
cology Center, Saint-Lucs Hospital, Montreal, Professor of Surgery,
University of Montreal.
11 11 Associate Professor of Surgery, Royal Victoria Hospital, Mon-
treal, Quebec, Canada.
llll Clinical Professor, Department of Surgery, Albert Einstein Col-
lege of Medicine of Yeshiva University, Bronx, NY.
## See Appendix 1.
Address for reprints: Bernard Fisher, MD, Professor of Surgery,
Department of Surgery, School of Medicine, University of Pittsburgh,
Pittsburgh, PA I5 I6 I .
Accepted for publication August 9, 1982.

0008-543X/83/1101/1551 $1.15 0 American Cancer Society

1551
1552 CANCERNovember I 1983 Vol. 5 2

r
80
loo r- Several recent NSABP prospective randomized clin-
ical trials2-' have provided an opportunity to reassess
the adequacy of the previously described positive nodal
groupings for determining the prognosis of breast cancer
; 6 0 ~
\
u
m
FIG. I . Basis for nodal grouping
patients and to expand upon the original findings. This
40 A- disease-free survival at 5 years. article presents information which continues to empha-
.\" size the importance of positive regional nodes as fore-
casters of distant disease. It reaffirms the effectiveness
of the previously established nodal groupings in that
regard, and it indicates that for greater prognostic pre-
cision there is need for further subdivision of those with
24 positive nodes.
positive axillary nodes were associated with a progres-
sively greater incidence of treatment failure.' A sharp Method
rise occurred when four nodes contained tumor. Con-
The 505 patients with primary breast cancer and his-
sequently, that trial, and all subsequent analyses of
tologically proven positive nodes utilized in the current
NSABP data, grouped patients with positive nodes into
analyses were participants in one or the other of two
those having 1 to 3 or 2 4 involved with tumor. The
prospective randomized clinical trials camed out by the
validity of such a grouping was confirmed by the ob-
NSABP. They were enrolled in those studies between
servance of a 25% difference in 5-year survival of pa-
197 1 and 1974; 336 in the first trial (B-04), which com-
tients in the 2 positive nodal groups. That categorization
pared the worth of radical to that of total mastectomy,
has become universally accepted and is utilized when
with or without postoperative irradiation, and 169 in
presenting findings regarding the natural history of
the second (B-03, which compared the disease-free sur-
breast cancer patients or the effectiveness of therapeutic
vival and the survival of patients receiving L-PAM ad-
regimens.
juvant therapy with those receiving placebo. Criteria of
patient eligibility, plans of investigation, entry infor-
TABLE1. Life Table Percent Probability of Disease-free mation as well as documentation of patient and tumor
Survival and Survival at 5 Years characteristics relative to each of' the trials, have been
r e p ~ r t e d . They
~ . ~ were essentially the same in both, as
Initial no. o f Disease-free
patients survival (9%) Survival (%)
were the operative technics employed. Consequently,
# Nodes patients with similar quantitative nodal involvement in
positive ALL 549 850 ALL 549 850 ALL 549 250 the two trials have been combined for clarity of presen-
0 296 88 208 84.9 81.7
tation. The findings from histologically negative node
(65) (171) patients came from those women entered into B-04,
1-3 249 84 I65 52.7 73.3
since B-05 was confined to patients with histologically
(37) (120) positive nodes. Data concerning the number of lymph
I Ill 40 71 57.3 78.8
nodes per specimen were derived from pathologic re-
(18) (54) ports submitted to the NSABP Biostatistical Center.
2 68 22 46 46.0 69.6
Actuarial life table methods have been employed to
(9) (33) summarize the distribution of the survival and disease-
3 70 22 48 5 I .o 68.8
free survival times.4 The number of patients at risk at
(10) (33) 5-years postmastectomy is shown to provide the reader
24 265 99 I66 25.9 48.8
with an idea of the reliability of the estimates at that
(21) (82) point. Statistical significance between survival curves is
4 53 21 32 33.3 53. I
based on a summary chi-square tesL5x6The hazard rates
(6) (17) were computed using life table methods.'
4-6 I09 39 70 41.8 55.7
(14) (39) Results
7-12 82 26 56 21.4 51.8
(4) (29) Disease-Free Survival at 5 Years Relative to Number
213 74 34 40 9.6 32.5
of Positive Nodes
(3) (14)
When the probability of all patients surviving disease
( ) Number of pattents at nsk. free was related to their axillary nodal status, a distinct
No. 9 NODESAND BREASTCA PROGNOSIS
POSITIVE

difference was observed between those with all negative

nodes and those with one positive node (Fig. 1). The
disease-free survival (DFS) of the former was 85% and
the latter 63%. There was little difference in DFS be-
tween those having one, two, or three nodes, with tumor.
A second abrupt decrease in DFS (from 57% to 40%)
was observed between patients with 3 positive nodes and
those with 4 nodes containing tumor. While little dif-
ference in DFS existed between those with four, five, or
six positive nodes, greater numbers of involved nodes
were associated with a progressively worse prognosis
(Table 1). The observance that all patients with 2 4 pos-
itive nodes had a 30.5% DFS, fails to indicate the di-
versity of findings within that nodal subgroup. Whereas
>
k
2 60
m
a
m
0
g
3
100

80

40

20

0 2
- Fisher et al.

PTS 5 4 9 Y R S

4 6
4-6

7-12

113

0 2 4
;
PTS ? 5 0 Y R S

6
1553

7-1;

patients having 4 nodes involved had a 39.7% DFS, only YEARS POSTMASTECTOMY

16.4% of those with 2 13 positive nodes were free of FIG. 2. Disease-free survival relative to age and number of positive
disease at that time. Such findings occurred both in pa- nodes for patients (PTS) 549 years and patients >=SOyears.
tients 5 4 9 years and those 250 years of age (Table l ,
Fig. 2). In the younger group there was a 32% difference
between those with 4 to 6 and those with 213 nodes during the second and third postoperative years. By the
containing tumor, whereas in the older patients the dif- fifth there was little difference between those having
ference was not as great (20%). negative, 1 to 3, or 14 positive nodes.
When the Hr was examined for subgroups of patients
Survival at 5 Years Relative to Number of within the 8 4 positive node group (Fig. 5, Table 3) it
Positive Nodes was observed that during the first few years following
When patients were examined overall, or according operation, the greater the number of nodes involved the
to age, findings relative to survival were concordant with greater was the probability of a patient having a treat-
those observed regarding DFS (Table 1, Fig. 3). While ment failure or dying. Those who completed the third
the conventional grouping of patients into negative, 1 year free of disease had a similar Hr whether they had
to 3 or 2 4 positive nodes, revealed 3 distinctive mor- 4 to 6, 7 to 12, or 2 13 positive nodes.
tality rates, subdivision of the 8 4 positive node group
indicated an increasing mortality associated with in- Discussion
creasing numbers of positive nodes. Whereas 47.1% of Despite universal acceptance of findings reported by
the patients with 4 nodes positive survived, when L 13 us in 1968' indicating the appropriateness of grouping
were involved only 28.4% of them were alive at that primary breast cancer patients according to their axillary
time. Similar findings were observed when patients were
examined according to age.
PTS 549 YRS PTS >50 YRS
- -
# #
Conditional Probability of Treatment Failure 100 POS POS.
3
or Mortality
0
0
80
The probability of a patient demonstrating a treat- 1-3 1-3
>
ment failure or dying during each successive year fol- e2
lowing operation, i.e., the hazard rate (Hr) was found m
60
4-6
to vary with the degree of nodal involvement (Fig. 4, a 4-6 7-12
m 7 - li
0
Table 2). In patients with no nodes positive, or with 1 40
to 3 positive, there was a constant conditional proba- ,\" 213
bility of failure in each group during this 5-year follow- 213

up. Those having no nodes demonstrated a lesser prob-

ability. When 2 4 positive nodes were present the Hr for
treatment failure was greatest during the first year after
operation, and decreased with each subsequent year,
2o
0 I
0 2 4 6
-
0 2 4
YEARS POSTMASTECTOMY
6
-

until at the fifth year it was the same as in those patients FIG. 3. Survival relative to age and number of positive nodes for
with 1 to 3 positive nodes. The mortality Hr was greatest patients (PTS) 549 years and patients 250 years.
I554 CANCERNovember I I983 Vol. 52

difference in the DFS (-20%) and survival (S) (1.10%)

FIG. 4. Conditional probability
according to negative, 1-3 or 2 4
0
positive nodes.
between those with negative axillary nodes and those
with I to 3 positive nodes. As previously noted, there
was relatively little difference in the outcome of patients
with one, two, or three positive nodes, but there was a
precipitous decrease in DFS and S when four nodes con-
tained tumor, reiterating the propriety of dividing pa-

ZZiJ
-MORTALITI -
tients into the two positive nodal groups. In the first
study a 25% difference in the 5-year survival was found
between those with 1 to 3 and those with 14 positive
1 2 3 4 5 nodes and in the recent evaluation the difference was
POSTOPERATIVE YEARS 27%. Thus, it has been shown in two independent in-
vestigations that it is more appropriate to classify pa-
tients as having l to 3 or 24 positive nodes than to
merely consider them as node negative or positive. The
major difference in DFS and S between those with 4 to
6 and those with 2 13 positive nodes indicates, however,
FIG. 5. Conditional probability that there is a risk in combining all patients with 24
according to number of positive positive nodes into a single group. Such a unification
nodes 2 4 . may provide misleading information regarding the prog-
nosis of patients, as well as the worth of therapeutic
regimens. Failure to ensure that treatment arms are bal-
anced, relative to the number of positive nodes possessed
-~ by patients in each group being compared, may result
POSTOPERATIVE YEARS
in ineffective therapies appearing effective and effectual
therapies being judged worthless. When comparing the
value of different treatments administered to a popu-
nodal status, i.e., negative, 1 to 3, or 2 4 positive, no lation of patients with 14 positive nodes should there,
systematic appraisal has been made to confirm or deny for example, be a preponderance with 4 to 6 nodes pos-
the suitability of that classification. The current results, itive in a group receiving one therapy and 213 nodes
obtained from more current NSABP protocols having positive in the group receiving another treatment, better
essentially the same eligibility criteria as in the study results occurring in the former group may be attributed
giving rise to our initial findings, completely support the to the therapy when they were in fact due to the greater
validity of the original observations. They indicate that number of patients with a better prognosis in that group.
5 years following operation there was a considerable It is no longer acceptable for authors indicating the ef-

TABLE
2. Disease-free Survival (Hazard rate X lo' f SEI

Years postmastectomy
# Positive Initial no.
nodes of patients 1 2 3 4 5

0 296 3.8 f 1.0 3.1 * 1.0 2.6 f 0.9 1.4 f 0.7 2.3 f 0.9
(294) (277) (259) (239) (22 1)
1-3 249 8.6 k 1.7 9.9 f 2.0 8.5 f 2.0 8.7 f 2.2 6.8 f 2.1
(245) (214) ( 1 84) (1 60) (139)
24 265 31.6 k 3.4 27.9 k 3.9 19.2 f 3.9 13.4 f 3.7 5.9 f 2.6
(260) (171) (1 16) (87) (73)
4-6 109 20.8 f 4.2 19.3 * 4.6 13.2 f 4.4 12.5 t 4.7 6.2 f 3.6
(108) (82) (61) (50) (42)
7-12 82 28.0 k 5.7 35.6 f 8.0 23.0 2 8.0 15.5 f 7.7 4.5 f 4.5
(80) (54) (33) (23) (19)
213 74 55.1 f 8.7 38.0 k 10.3 32.4 f 12.0 12.8 f 9.0 7.2 f 7.2
(72) (35) (21) (14) (12)

( ) Number of patients at risk.

No. 9 NODESAND BREASTCA PROGNOSIS Fisher ef al.
POSITIVE . 1555

TABLE3. Survival (Hazard rate X lo f SE)

Years postmastectomy
# Positive Initial no.
nodes of patients 1 2 3 4 5

0 296 2.3 f 0.8 1.7 f 0.7 4.5 f 1.2 3.2 ? 1.0 4.0 f 1.1
(296) (288) (282) (267) (257)
1-3 249 2.0 f 0.8 4.6 f 1.3 6.0 f 1.5 6.9 f 1.7 6.7 ? 1.7
(249) (243) (230) (2 I 3) (194)
24 265 9.3 f 1.7 19.2 f 2.7 17.6 ? 2.9 10.5 f 2.5 8.5 f 2.3
(265) (237) ( 1 88) ( 152) ( 134)
4-6 I09 4.7 f 1.9 15.0 f 3.6 18.3 k 4.4 6.3 f 2.8 6.8 f 3.0
(109) (103) (86) (69) (64)
7-12 82 9.7 f 3.2 19.1 f 4.9 9.1 f 3.7 15.8 f 5.2 4.0 f 2.8
(82) (73) (58) (52) (43)
213 74 16.0 f 4.4 27.0 f 6.5 28.9 f 7.9 11.5 f 5.7 20.8 f 8.4
(74) (61) (44) (31) (27)

( ) Number of patients at risk.

fectivenessof one therapy over another to assume equiv-
alence of the patients in the treatment groups being com-
pared because the proportion of node-positive patients
was similar, the average number of positive nodes was
equivalent, or the proportion of 2 4 positive-node
patients was almost identical. Such statements, or even
the data which support them, fail to ensure concordance
of the populations under consideration.
Although the current studies portray the relationship
between the extent of the nodal involvement of breast
cancer patients and their markedly heterogeneous nat-
ural history, they fail to do so completely. They do not
compare the outcome of patients with extranodal ex-
tension of metastases with those having metastatic de-
posits within the confines of the nodal capsule. In ad-
dition, they provide no information relative to the fate
of patients with nodal micro- and macrometastases, nor
do they indicate the natural history of patients with oc-
cult axillary metastases. Previous publications from the
NSABP have, however, addressed themselves to those
facets of nodal involvement and patient
It has been reported* that when extranodal extension of
tumor occurs, the treatment failure rate is almost twice
that observed when tumor is confined to the node. An-
other study has noted that there was no difference in
survival between patients with negative nodes and those
with micrometastases (<2 mm). Both of those groups
exhibited a significantly greater survival than patients
with macrometastases ( 2 2 mm). Treatment failure rates
in patients with micrometastases were more comparable
to those with macrometastases than to those with absent
nodal involvement. Further evaluation (multivariate
analysis) indicates, however, that the findings relative to
both survival and treatment failure were more directly
related to the number of nodes involved than with the
size of tumor in the node. One subset of patients with
micrometastases, i.e., those measuring S 1.3 mm, ex-
hibited survival and treatment failure rates similar to
those of negative node patients and the findings were
independent of the number of nodes involved. That
observation coincided with another indicating that the
survival of patients with negative nodes was similar
whether or not those nodes were found, as a result of
serial section, to contain occult tumor.O It is thus ap-
parent that factors other than numbers of nodes in-
volved with tumor may require consideration when re-
lating nodal status to prognosis, and especially when
determining the equivalence of patient populations.
Of particular interest are the findings relating the con-
ditional probability of a treatment failure or death each
year during the 5-year period following operation to the
degree of nodal involvement. They indicate that for
those with negative or 1 to 3 positive nodes the risk of
developing a treatment failure or dying was relatively
constant during that period, whereas in those having
2 4 positive nodes the probability was highest during the
first few years but by the fifth year the risk was the same
as for the group having 1 to 3 positive nodes. In the
2 4 positive node group the risk in the early years was
directly related to the number of nodes positive, but by
the fourth year the risk was the same whether 4 to 6 or
2 13 nodes were involved. These findings, particularly
for those with 14 nodes positive, support our contention
that systemic adjuvant therapy, which fails to demon-
strate a positive effect by two years following operation,
is unlikely to do so eventually, since so many patients
have already failed or died of their disease. Those who
do survive free of disease beyond that time are at de-
creasing risk for recurrence. Based on Hrs, we have con-
sidered it justifiable to report preliminary results re-
1556 CANCERNovember I
garding the effectiveness of a therapeutic regimen after
two years of observation.
Recent reports have indicated that tumor estrogen
receptor (ER) levels may be an important discriminant
of progno~is.''-'~ It has been suggested that their pre-
dictive value may not only be independent of, but may
be of greater importance than nodal status. Knight and
associates have demonstrated' ' that 18 months follow-
ing operation patients with ER-negative tumors have
about a 2- to 3-fold higher recurrence rate than do those
with ER-positive tumors. That difference between those
with ER negative and positive tumors existed in patients
with negative, 1 to 3, or 1 4 positive axillary nodes.
When examined relative to nodal status, independent
of tumor ER level, a 5-fold difference in recurrence ex-
isted between those with negative and 24 positive nodes
suggesting that although both ER and nodal status are
of prognostic importance, nodal status remains the pre-
dominant indicator of patient outcome. In NSABP stud-
ies when multivariate analysis was employed, of all of
the putative prognosticators, e.g., age, size of primary
tumor, histologic and nuclear grade, the dominant van-
able was found to be nodal status. The findings of Hah-
nel et a1.,I2further indicate the primacy of nodal infor-
mation for estimating prognosis. While their results in
the first two years after operation are in agreement with
those of Knight et al.," by 5 years the probability of
recurrence was found to be almost identical for patients
with the same nodal status having ER-positive or neg-
ative tumors. In contrast, as with our findings, differ-
ences in patient outcome related to nodal status contin-
ued to persist at 5 years. I t is of interest that Hilf and
coworker^,'^ have failed to demonstrate the usefulness
of ER status as a prognosticator or recurrence, and Rich
and associate^'^ have found that the presence of ER
cannot be correlated with lymph node tumor infiltra-
tion.
It has been repeatedly emphasized by us that the prob-
ability of development of distant disease and mortality
is unrelated to the degree of axillary dissection at the
time of ma~tectorny.'~.'~ The primary purpose of such
a dissection is to obtain information regarding the his-
tologic nodal status of the patient. That information can
be utilized for estimating prognosis and for determining
the need for systemic adjuvant therapy. Should tumor
ER be equal to, or more sensitive than, nodal status as
a discriminant of prognosis or as an indicator of patients
who should receive chemotherapy, another justification
for axillary dissection would be eliminated. For the pres-
ent, at least, there is no evidence to justify utilization
of ER in lieu of nodal status; both should be employed
in concert.
Since there is need to quantify the degree of positive
nodal involvement, more than an axillary sampling is
1983 Vol. 52
indicated. We have previously demonstrated16that when
only a few nodes are removed the qualitative axillary
nodal status, that is, positive or negative, can be deter-
mined with accuracy. When, however, 5 10 nodes are
removed, the greater is the likelihood that some of the
patients having 2 4 nodes positive will be considered to
have 1 to 3 positive nodes. Thus, while removal of more
nodes does not influence patient outcome, it does ensure
more accurate assessment of the degree of positive node
involvement. Our previous findings have indicated that
removal of the lower two levels of axillary nodes is ad-
equate to fulfill that need.
REFERENCES
I. Fisher B, Ravdin RG, Ausman RK, Slack NH, Moore GE, Noer
RJ, Cooperating Investigators. Surgical adjuvant chemotherapy in can-
cer of the breast: Results of a decade of cooperative investigation. Ann
Surg 1968; 168:337-356.
2. Fisher B, Montague E, Redmond C, et al., Other NSABP In-
vestigators. Comparison of radical mastectomy with alternative treat-
ments for primary breast cancer: A first report of results from a pro-
spective randomized clinical trial. Cancer 1977; 39:2827-2839.
3. Fisher B, Carbone P, Economou SG et al.. Other Cooperating
Investigators. L-phenylalanine mustard (L-PAM) in the management
of primary breast cancer: A report of early findings. N Engl J Med
1975; 29211 17-122.
4. Cutler SJ, Ederer F. Maximum utilization ofthe life table method
in analyzing survival. J Chronic Dis 1958; 8:699-7 12.
5 . Mantel N. Evaluation of survival data and two new rank order
statistics arising in its consideration. Cancer Chemuther 1966; 50: 163-
170.
6. Pet0 R, Pet0 J. Asymptotically efficient rank invariant test pro-
cedures. J R Stat Suc 1972; A 135: 185-206.
7. Lee ET. Statistical Methods for Survival Data Analysis. Belmont,
CA: Lifetime Learning Publications, 1980.
8. Fisher ER, Gregorio RM, Redmond C, Kim WS, Fisher B.
Pathologic findings from the National Surgical Adjuvant Breast Project
(Protocol No. 4) HI: The significance of extranodal extension of ax-
illary metastases. Am J CIin Pathol 1976; 65:439-444.
9. Fisher ER, Palekar A, Rockette H, Redmond C, Fisher B. Patho-
logic findings from the National Surgical Adjuvant Breast Project
(Protocol No. 4) V: Significance of axillary nodal micro- and macro-
metastases. Cancer 1978; 42:2032-2038.
10. Fisher ER, Swamidoss S, Lee CH, Rockette H, Redmond C,
Fisher B. Detection and significance of occult axillary node metastases
in patients invasive breast cancer. Cancer 1978; 42:2025-203 I .
I I. Knight WA, Livingston RB, Gregory EJ, McGuire WL. Estro-
gen receptor as an independent prognostic factor for early recurrence
in breast cancer. Cancer Res 1977; 37:4669-467 I .
12. Hiihnel R, Woodings T, Vivian AB. Prognostic value of estro-
gen receptors in primary breast cancer. Cancer 1979; 44:67 1-675.
13. Rich MA, Furmanski P, Brooks SC, the Breast Cancer Prog-
nostic Study Surgery and Pathology Associates. Prognostic value of
estrogen receptor determinations in patients with breast cancer. Cancer
Res 1978; 38:4296-4298.
14. Hilf R, Feldstein ML, Gibson SL, Savlov ED. The relative im-
portance of estrogen receptor analysis as a prognostic factor for re-
currence or response to chemotherapy in women with breast cancer.
Cancer 1980; 45:1993-2OOO.
15. Fisher B. Breast cancer management: Alternatives to radical
mastectomy. N Engl J Med 1979; 301:326-328.
16. Fisher B, Wolmark N, Bauer M, Redmond C, Gebhardt M.
The accuracy of clinical nodal staging and of limited axillary dissection
as a determinant of histologic nodal status in carcinoma of the breast.
Surg Gynecol Obstet 198 1; 152:765-772.
No. 9 POSITIVE NODESAND BREASTCA PROGNOSIS - Fisher et a/. 1557

APPENDIX 1. NSABP INSTITUTIONS AND PRINCIPAL INVESTIGATORS PARTICIPATING IN PROTOCOLS B-04 AND B-05

Institution Principal Investigator

Akron City Hospital Marvin J. Sakol

Boston City Hospital C. William Kaiser
Cleveland Metropolitan Hospital Edward G. Mansour
Creighton University Claude H. Organ
Denver General Hospital George E. Moore
Duke University Center William Shingleton
Ellis Fischel State Cancer Hospital William Donegan
Fitzsimons General Hospital Richard M. Hirata
French and Polyclinic Medical School James McManus
Geisinger Medical Center C. W. Konvolinka
Georgia Baptist Hospital A. Hamblin Letton
Greater Bakersfield Memorial Hospital James F. Donovan
Group Health Medical Center Robert Bourdeau
Harbor General Hospital, UCLA John R. Benfield
Harrison S. Martland Hospital Benjamin F. Rush Jr.
Hennepin County General Hospital Claude R. Hitchcock
Hershey Medical Center William E. DeMuth
Highland Hospital (Rochester, NY) Edwin Savlov
Hotel-Dieu, Quebec City Louis Dionne
Huron Road Hospital M. D. Ram
Kaiser Permanente, Portland Andrew Glass
Letterman General Hospital Harvey Conklin/Michael Corder
Louisiana State University Isidore Cohn Jr.
Lynn Hospital (Lynn, MA) Bernard Willett
Medical College of Pennsylvania Donald Cooper
Medical College of Virginia Walter Lawrence Jr.
Memorial Cancer Research Foundation of Southern California David Plotkin/Elliott Hinkes
Mercy Hospital (Des Moines, IA) Joseph Song
Metropolitan Hospital John F. Weiksnar
Michael Reese Hospital Richard H. Evans
Montefiore Hospital Richard G. Rosen
Montreal General Hospital John MacFarlane
Mount Sinai, New York Gerson J. Lesnick
Newark Beth Israel Frederick B. Cohen
Ohio State University John P. Minton
Roger Williams General Hospital (Providence, RI) Jack Savran
Rush Presbyterian-St. Lukes Medical Center Steven Economou/Charles Perlia
St. Lukes Hospital (Kansas City, MO) Paul Koontz
St. Vincents Hospital, New York Thomas Nealon Jr.
Southern California Permanente (San Diego, CA) Thomas Campbell
Temple University Willis Maier
Tufts-Pondville Hospital (Walpole, MA) Leo Stolbach
University of Alberta Pierre Band
University of Arkansas Kent Westbrook
University of California, San Diego Marshall J. Orloff
University of Florida, Jacksonville Neil Abramson
University of Illinois Tapas K. Das Gupta
University of Iowa Richard L. Lawton
University of Louisville Condict Moore/Carl Knutson
University of Mississippi George V. Smith
University of Rochester W. Bradford Patterson
U. S. Naval Hospital, San Diego T. James Guzik
USC-LA County Medical Center Lewis Guiss
University of Pennsylvania Francis E. Rosato
University of Texas, Galveston Edward B. Rowe
Washington University George J. Hill/Harvey Butcher
Wayne State University Alexander J. Walt
West Suburban Hospital (Chicago, IL) Everett Nicholas
White Memorial Medical Center Samuel Fritz
Wilmington Medical Center Robert W. Frelick