Original Article

The Effect of Paget Disease on Axillary Lymph Node

Metastases and Survival in Invasive Ductal Carcinoma
Stephanie M. Wong, MD1,2; Rachel A. Freedman, MD, MPH3; Yasuaki Sagara, MD4; Emily F. Stamell, MD1,3;
Stephen D. Desantis, BS3,4; William T. Barry, PhD5; and Mehra Golshan, MD4

BACKGROUND: The objective of this study was to examine the effect of Paget disease (PD) on axillary lymph node metastases and
survival in patients who had concomitant invasive ductal carcinoma (PD-IDC). METHODS: The Surveillance, Epidemiology, and End
Results (SEER) database was used to identify women who were diagnosed with PD-IDC from 2000 to 2011, comparing baseline de-
mographic and tumor characteristics with those who were diagnosed with IDC alone during the same period. Multivariable logistic
regression was used to examine the association of PD-IDC with axillary lymph node metastasis, and breast cancer-specific survival
and overall survival were compared between the PD-IDC and IDC groups using the Kaplan-Meier method and Cox proportional haz-
ards regression. RESULTS: The study cohort included 1102 patients with PD-IDC and 302,242 controls with IDC alone. PD-IDC tumors
were more likely to be centrally located (26.9% vs 5.5%; P <.001), high grade (63.5% vs 40.3%; P <.001), >2 cm in greatest dimension
(47.1% vs 35.7%; P <.001), and estrogen/progesterone receptor-negative (45.2% vs 22.1%; P <.001). In adjusted analyses, patients with
PD-IDC had higher odds of axillary lymph node metastasis (odds ratio, 1.83; P <.001). The unadjusted 10-year breast cancer-specific
and overall survival rates were lower for the PD-IDC group compared with the IDC-alone group, although, after adjusting for disease
stage, tumor characteristics, and local therapy, no significant differences in mortality risk were observed between the 2 groups (haz-
ard ratio, 0.91; P 5.24). CONCLUSIONS: PD-IDC is associated with an increased risk of axillary lymph node metastasis, but not with in-
ferior survival, compared with IDC alone after adjustment for other disease factors. Cancer 2015;121:4333-40. V C 2015 American

Cancer Society.

KEYWORDS: breast neoplasms, cancer outcomes, Paget disease of the breast, prognosis, Surveillance, Epidemiology, and End
Results, survival.

INTRODUCTION
Paget disease (PD) of the nipple is a rare clinical manifestation of breast carcinoma; it is present in an estimated 0.5% to
3% of newly diagnosed patients1,2 and is characterized by neoplastic infiltration of large, pale, Paget cells into the epider-
mis of the nipple-areolar complex.3 Although the mechanism underlying this entity is not well understood, 2 theories
have been proposed to explain the pathogenesis of PD. The epidermatropic theory suggests that cells arise from an underly-
ing ductal tumor and migrate along the lactiferous ducts to reach the nipple epidermis, where symptoms manifest.4,5
In contrast, the transformation theory posits that malignant cells arise in situ from the transformation of keratinocytes
within the major lactiferous sinuses.6
In the majority of patients, PD is associated with an underlying in situ or invasive malignancy, with concurrent inva-
sive ductal carcinoma (IDC) in 50% to 60% of patients and in situ disease in approximately 30% to 40%.5,7,8 Further-
more, the corresponding IDC pathology has been characterized as larger tumor size, higher grade, and more likely to be
negative for both estrogen receptor (ER) and progesterone receptor (PR) (hormone receptor-negative) and positive for
human epidermal growth factor receptor 2 (HER2/neu).7,9,10
Recent literature suggests that the presence of PD may be associated with poor prognostic outcomes and reduced sur-
vival.11-13 However, because PD often presents with axillary lymph node involvement and other prognostic factors for
poor outcomes, it is difficult to discern from small studies whether PD itself confers an increased risk of mortality. In this

Corresponding author: Mehra Golshan, MD, Department of Surgery, Dana-Farber Cancer Institute and Brigham and Women’s Hospital, 450 Brookline Avenue,
Boston, MA 02115; Fax: (617) 582-7740; mgolshan@partners.org
1
Harvard School of Public Health, Boston, Massachusetts; 2Department of Surgery, McGill University Health Center, Montreal, Quebec, Canada; 3Department of
Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts; 4Department of Surgery, Brigham and Women’s Hospital/Dana-Farber Cancer Institute,
Boston, Massachusetts; 5Department of Biostatistics and Computational Biology, Dana-Farber Cancer Institute, Boston, Massachusetts

Dr. Freedman acknowledges the American Cancer Society and the Susan G. Komen for the Cure Foundation, but has not received any funding support from
these bodies in relation to the current study.

DOI: 10.1002/cncr.29687, Received: July 13, 2015; Revised: August 14, 2015; Accepted: August 24, 2015, Published online September 16, 2015 in Wiley Online
Library (wileyonlinelibrary.com)

Cancer December 15, 2015 4333

Original Article
Figure 1. Stepwise cohort ascertainment is illustrated for the
current study.
study, we used population-based data to examine whether
the presence of PD was associated with an increased risk
of axillary lymph node metastases in patients with an
underlying IDC (PD-IDC). We also examined the associ-
ation between PD and survival.
MATERIALS AND METHODS
Data Source
We used data from the National Cancer Institute’s Sur-
veillance, Epidemiology, and End Results (SEER) pro-
gram, which contains the population-based central cancer
registries of 18 geographically defined regions that, to-
gether, represent approximately 28% of the US popula-
tion. For this study, we used the November 2013
SEER-18 submission, which contains patients from the
following geographic regions: Metropolitan Atlanta, Con-
necticut, Detroit, Hawaii, Iowa, New Mexico, San
Francisco-Oakland, Seattle-Puget Sound, Los Angeles,
San Jose-Monterey, Utah, Rural and Greater Georgia,
Alaska, Greater California, Kentucky, Louisiana, and
New Jersey. Because the study used publically available
data with no personal identifiers, the protocol was consid-
ered exempt from the Institutional Review Board of the
Dana-Farber Cancer Institute.
Patient Selection
Histopathology codes from the International Classification
of Diseases for Oncology third edition (ICD-O-3) were
used to identify patients who had PD and underlying
IDC (ICD-O-3 code 8541/3) as well as patients who had
IDC alone (ICD-O-3 code 8500/3). In our analyses, we
4334
included women who were diagnosed between 2000 and
2011 with histologically confirmed stage I through III dis-
ease and had no history of any prior cancer (n 5 1426).
Patients who had ER or PR status that was borderline or
unknown (n 5 187) were excluded from the analysis
along with those who lacked complete information related
to tumor size and histologic grade (n 5 86). Finally, we
excluded those who did not undergo surgery with axillary
lymph node evaluation (n 5 51). Thus, the final cohort
consisted of 1102 women who had PD-IDC and, apply-
ing identical inclusion and exclusion criteria, a control
group of 302,242 women who had IDC alone (Fig. 1).
Outcome of Interest
Our outcome of interest was the presence of axillary lymph
node metastases, which we defined as positive if 1 or more
lymph nodes were reported as positive on pathologic exam-
ination. To quantify the extent of axillary lymph node
involvement, we obtained the total number of axillary
lymph nodes removed and the total number of positive
lymph nodes. Then, we compared the total number of
positive lymph nodes across all patients (Table 1) as well as
exclusively in those who were deemed to have undergone
an axillary lymph node dissection (ALND) (Fig. 2).
Because SEER does not code for ALND specifically, simi-
lar to other studies, we defined receipt of ALND as having
>5 lymph nodes excised, which meets the American Joint
Committee on Cancer definition of a low ALND.14,15 A
secondary outcome of interest was survival. We calculated
the length of breast cancer-specific survival (BCSS) and
overall survival (OS) as the number of months between the
date of diagnosis and the last date for which vital status was
available. All patients who remained alive on December
31, 2011, were considered censored.
Independent Variables of Interest
Additional variables of interest included age at diagnosis
(with patients divided into groups aged <40, 40-59, 60-
79, and 80 years), race/ethnicity (white, black, and His-
panic/other/unknown), tumor size (0-2.0 cm, 2.1-5 cm,
and >5 cm), histologic grade (grade 1, 2, and 3/4), tumor
location (reported as central/nipple-areolar complex or by
quadrant), hormone receptor status (combining ER and
PR status to yield 4 subgroups; ER-positive/PR-positive,
ER-positive/PR-negative, ER-negative/PR-positive, and
ER-negative/PR-negative), type of surgery (breast-con-
serving surgery, mastectomy), and receipt of radiation
therapy. For the 18.4% of patients who had HER2/neu
status available, disease was defined as positive, negative,
or borderline/unknown, as defined by the Derived HER2
Summary variable from SEER.
Cancer December 15, 2015
 Survival in Paget Disease of the Breast/Wong et al

TABLE 1. Patient Characteristics

No. of Patients (%)

Characteristic PD-IDC, N 5 1102 IDC, N 5 302,242 P

Age at diagnosis : Mean [95% Cl], y 58.9 (58.0-59.8) 58.6 (58.6-58.7) 0.46
Race 0.12
White 870 (79.0) 245,848 (81.3)
Black 125 (11.3) 29,683 (9.8)
Hispanic/Other/Unknown 107 (9.7) 26,711 (8.8)
Tumor location <0.0001
Central/NAC 296 (26.9) 16,589 (5.5)
Upper Inner Quadrant 47 (4.3) 35,778 (11.8)
Lower Inner Quadrant 50 (4.5) 18,007 (6.0)
Upper Outer Quadrant 166 (15.1) 110,885 (36.7)
Lower Outer Quadrant 53 (4.8) 21,990 (7.3)
Overlapping lesion/Breast NOS 490 (44.5) 98,993 (32.8)
Tumor grade <0.0001
Grade I 70 (6.4) 56750 (18.8)
Grade II 332 (30.1) 123634 (40.9)
Grade III 700 (63.5) 121858 (40.3)
Tumor size, cm <0.0001
0 – 2.0 cm 583 (52.9) 194,411 (64.3)
2.1 – 5.0 cm 406 (36.8) 92,650 (30.7)
>5.0 cm 113 (10.3) 15,181 (5.0)
Hormone receptor status <0.0001
ER1/PR1 379 (34.4) 195,697 (64.7)
ER1/PR- 184 (16.7) 35,611 (11.8)
ER-/PR1 41 (3.7) 4,124 (1.4)
ER-/PR- 498 (45.2) 66,810 (22.1)
HER2 status* <0.0001
Positive 92 (59.4) 9,317 (15.9)
Negative 59 (39.1) 47,689 (81.5)
Borderline 4 (2.6) 1,509 (2.6)
Surgery <0.0001
Breast conserving surgery 97 (8.8) 177,715 (58.8)
Mastectomy 1005 (91.2) 124,527 (41.2)
Lymph Node status
Total Positive 584 (53.0) 103,030 (34.1) <0.0001
1-3 positive lymph nodes 340 (30.9) 71193 (23.6)
4-9 positive lymph nodes 152 (13.8) 21587 (7.1)
10 positive lymph nodes 82 (7.4) 8985 (3.0)
Node positive, NOS 10 (0.9) 1265 (0.4)
Total Negative 517 (47.0) 199,016 (65.9)
Unknown 1 (0.001) 196 (0.001)

Abbreviations: 2, negative; 1, positive; CI, confidence interval; ER, estrogen receptor; HER2, human epidermal receptor-2; IDC, invasive ductal carcinoma;
NAC, nipple areola complex; NOS, not otherwise specified; PD-IDC, Paget disease with invasive ductal carcinoma; PR, progesterone receptor.
a
HER2 status was available only for patients who were diagnosed during 2010 through 2011.

Statistical Analysis We performed survival analyses using the Kaplan-

Baseline patient demographic characteristics and tumor in-
formation were compared using the Student t test for con-
tinuous data and the Pearson chi-square test for categorical
variables. We then constructed a multivariable logistic
regression model to assess the probability of axillary lymph
node metastasis in the PD-IDC and IDC groups. The
model was adjusted for age, tumor location, histologic
grade, tumor size, and combined hormone receptor (ER
and PR) status. Next, we performed a subgroup analysis on
women who were diagnosed during the years 2010 and
2011, when information on HER2/neu status first became
available through SEER. This model included all prespeci-
fied variables as well as HER2/neu receptor status.
Meier method and compared the distribution of BCSS
and OS between the PD-IDC and IDC groups using the
log-rank test. Then, we constructed a multivariable Cox
proportional hazards model to evaluate the effect of PD
on survival after adjusting for age, lymph node status, tu-
mor size, histologic grade, hormone receptor status, and
type of local therapy (breast-conserving surgery or mastec-
tomy, receipt of radiation). The proportional hazards
assumption was verified. Survival functions were then
reestimated using the Kaplan-Meier method after weight-
ing by inverse propensity score to balance groups accord-
ing to all variables that were present within the
multivariable model.

Cancer December 15, 2015 4335

Original Article

patients in the PD-IDC group were more likely to have

Figure 2. This chart illustrates the percentages of patients
with invasive ductal carcinoma (IDC) alone and those with
Paget disease (PD) plus IDC (PD-IDC) who had lymph node-
positive disease and had >5 lymph nodes examined. Patients
with PD-IDC had significantly greater lymph node involve-
ment across lymph node classification groups (P <.001). N1
indicates all lymph nodes positive and any number of axillary
lymph node metastases (56.1% IDC vs 68.5% PD-IDC); N1, me-
tastases in 1 to 3 axillary lymph nodes (36.5% IDC vs 37% PD-
IDC); N2, metastases in 4 to 9 axillary lymph nodes (13.75%
IDC vs 20.2% PD-IDC); N3, metastases in 10 axillary lymph
nodes (5.9% IDC vs 11.3% PD-IDC).
All analyses were conducted using SAS software ver-
sion 9.4 (SAS Institute Inc, Cary, NC). All statistical tests
were 2-sided, and P values < .05 were considered statisti-
cally significant, and all confidence intervals (CI) are
stated at the 95% confidence level.
RESULTS
Cohort Characteristics
The 1102 patients who had PD-IDC and the 302,242
patients who had IDC alone were similar with respect to
demographic indices, such as mean age (58.9 vs 58.6
years, respectively; P 5 .46) and racial distribution
(P 5 .12). However, with respect to tumor characteristics,
TABLE 2. Distribution of Breast Subtype,
n 5 55,710a
Breast Cancer PD-IDC, IDC,
Subtypeb N 5 143 N 5 55,567
HER21/HR2 40 (28) 2805 (5.1)
HER21/HR1 47 (32.9) 6127 (11)
HER22/HR1 49 (34.2) 39,127 (70.4)
Triple negative 7 (4.9) 7508 (13.5)
Abbreviations: 2, negative; 1, positive; ER, estrogen receptor; HER2,
human epidermal receptor-2; HR, hormone receptor (estrogen and proges-
terone receptor); IDC, invasive ductal carcinoma; NAC, nipple areola com-
plex; NOS, not otherwise specified; PD-IDC, Paget disease with invasive
ductal carcinoma; PR, progesterone receptor.
a
Subtype data were available only for patients who were diagnosed during
2010 through 2011.
b
Patients who had borderline HER2 status were excluded.
centrally located tumors (26.9% vs 5.5%; P < .001),
lesions of higher histologic grade (63.5% vs 40.3% had
grade 3 or 4 lesions; P < .001) and larger size (47.1% vs
35.7% had tumors >2 cm; P < .001) compared with the
IDC-alone group (Table 1).
Hormone expression profiles varied significantly
between groups. More patients with PD-IDC were hor-
mone receptor-negative (45.2%) compared with only
22.1% of patients in the IDC-alone group (P < .001).
For those diagnosed during 2010 and 2011 who had hor-
mone receptor and HER2/neu status available
(n 5 55,710; 18.4% of the cohort), 60.8% of patients in
the PD-IDC group were HER2/neu–positive compared
with only 16.1% of the IDC-alone group (P < .001). The
distribution of breast subtype in the PD-IDC and IDC-
alone groups is provided in Table 2.
Risk of Axillary Lymph Node Metastasis
Overall, 53% of patients in the PD-IDC group were posi-
tive for lymph node metastasis compared with 34.1% of
patients in the IDC-alone group (P < .001). Figure 2
illustrates the distribution of axillary lymph node involve-
ment among the patients who had a total of >5 lymph
nodes removed, indicating a higher proportion of lymph
node-positive disease among those who had PD present
(P < .001).
In the adjusted analysis assessing for the presence of ax-
illary lymph node metastasis (Table 3), PD remained a sig-
nificant predictor of axillary lymph node metastasis after
controlling for age and tumor characteristics, such as histo-
logic grade, location, tumor size, and hormone receptor sta-
tus (odds ratio, 1.83; 95% CI, 1.61-2.08). We also
examined the sensitivity of these findings to an additional
potential confounder: HER2/neu amplification. By using
patients for whom HER2/neu status was available (those
diagnosed between 2010 and 2011), we performed a sub-
group analysis adjusting for HER2/neu amplification as well
as the variables previously described and observed that the
P presence of PD remained a statistically significant predictor
of axillary metastasis (odds ratio, 1.76; 95% CI, 1.23-2.51).
< .001
Effect of PD on Survival Outcomes
The median length of follow-up was 55 months for the
IDC group (interquartile range, 26-91 months) and 55.5
months for the PD-IDC group (interquartile range,
28-91 months). Overall, patients in the PD-IDC group
had a 5-year BCSS of 85.1% compared with 92.4% in the
IDC-alone group (absolute difference, 7.3%; log-rank
test, P < .001). The 10-year BCSS rate was 77.8% for the

4336 Cancer December 15, 2015

 Survival in Paget Disease of the Breast/Wong et al

TABLE 3. Logistic Regression Analysis for Axillary Lymph Node Metastasis Among Patients Diagnosed With
Invasive Ductal Carcinoma With and Without Paget Disease (n 5 303,147)a

% With Axillary P Adjusted OR for

Lymph Node Axillary Lymph Node
Characteristic Total No. (%) Metastasis P Metastasis [95% CI]

Histology
IDC only 302,046 (99.6) 34.1 < .0001 1.00
PD-IDC 1101 (0.36) 53 1.83 [1.61-2.08]b
Age group, y
<40 19,709 (6.5) 48.6 < .0001 1.00
40-59 143,761 (47.4) 37.2 0.79 [0.77-0.82]b
60-79 119,372 (39.4) 28.51 0.59 [0.57-0.61]b
>80 20,305 (6.3) 32.1 0.60 [0.58-0.61]b
Tumor location
Central/NAC 16,874 (5.6) 42 < .0001 1.00
Upper inner quadrant 35,800 (11.8) 22.8 0.44 [0.42-0.46]b
Lower inner quadrant 18,040 (6) 27.2 0.59 [0.57-0.62]b
Upper outer quadrant 110,993 (36.6) 35.7 0.82 [0.79-0.85]b
Lower outer quadrant 22,030 (7.3) 36.2 0.86 [0.82-0.90]b
Overlapping quadrants/breast NOS 99,410 (32.8) 36.2 0.80 [0.77-0.83]b
Tumor grade
1 56,796 (18.7) 19.1 < .0001 1.00
2 123,870 (40.9) 32.6 1.67 [1.62-1.71]b
3 52,368 (40.4) 42.8 2.08 [2.02-2.14]b
Tumor size, cm
0.0-2.0 194,882 (64.3) 22.6 < .0001 1.00
2.1-5.0 92,993 (30.7) 52.3 3.35 [3.29-3.40]b
>5.0 15,272 (5) 72.1 7.58 [7.29-7.87]b
Hormone receptor status
ER1/PR1 195,955 (64.6) 32.8 < .0001 1.00 [Referent]
ER1/PR2 35,779 (11.8) 35.1 0.92 [0.90-0.92]b
ER2/PR1 4163 (1.37) 36.8 0.72 [0.69-0.82]b
ER2/PR2 67,250 (22.2) 37.7 0.76 [0.70-0.80]b

Abbreviations: 2, negative; 1, positive; ER, estrogen receptor; IDC, invasive ductal carcinoma; NAC, nipple areola complex; NOS, not otherwise specified;
PD-IDC, Paget disease with invasive ductal carcinoma; PR, progesterone receptor.
a
Patients with unknown pathologic lymph node status were excluded from the analysis.
b
After adjusting for all other variables in the table, these values were statistically significant (P <.05).

PD-IDC group and 87% for the IDC-alone group (abso-
lute difference, 9.2%; log-rank test, P < .001) (Fig. 3A).
The 5-year and 10-year OS rates were 77.9% and
61.8%, respectively, in the PD-IDC group compared
with 87.5% and 74.9%, respectively, in the IDC-alone
group (absolute difference: 5-year OS, 9.6%; 10-year OS,
13.1%; log-rank test, P < .001) (Fig. 3C).
On unadjusted Cox regression analysis, patients
in the PD-IDC group demonstrated statistically
worse BCSS compared with the IDC-alone reference
group (hazard ratio, 1.92; 95% CI, 1.63-2.25).
However, on multivariable analysis adjusting for age,
lymph node status, tumor size, histologic grade, hor-
mone receptor status, type of surgery, and receipt of
radiation, patients with PD-IDC had a risk of mor-
tality similar to that of patients in the IDC-alone
group (hazard ratio, 0.91; 95% CI, 0.77-1.07;
P 5 .24) (Table 4). Weighted Kaplan-Meier curves
for BCSS and OS further demonstrated no signifi-
cant difference between the 2 groups after balancing
for tumor characteristics and treatment variables
according to propensity scores (Fig. 3B,D).
DISCUSSION
In this large, population-based cohort, >50% of women
with PD-IDC had axillary lymph node involvement,
compared with only 34.1% of patients with standard
IDC. After controlling for tumor characteristics known to
predict lymph node involvement, the presence of PD
remained independently associated with axillary metasta-
sis. Although the unadjusted 5-year and 10-year survival
rates differed for patients with IDC and PD, these find-
ings did not persist in adjusted analyses, in which we
observed similar hazard rates between the 2 groups after
adjusting for tumor and treatment variables. These find-
ings suggest that, although PD is associated with more
advanced disease at presentation, it does not by itself con-
fer a poorer prognosis in women with IDC.
Previous literature has emphasized the biologically
more aggressive nature of tumors associated with PD.

Cancer December 15, 2015 4337

Original Article

Figure 3. Kaplan-Meier curves compare survival between patients who had invasive ductal carcinoma (IDC) and those who had
Paget disease (PD) with IDC (PD-IDC). (A) Unweighted and (B) inverse propensity score (IPS)-weighted breast cancer-specific
survival is illustrated for all patients with stage I through III PD-IDC and for patients with IDC alone at 5 years and 10 years. (C)
Unweighted and (D) IPS-weighted overall survival is illustrated for all patients with stage I through III PD-IDC and for patients
with IDC alone at 5 years and 10 years.

Several retrospective studies and institutional case series
have demonstrated high rates of axillary lymph node me-
tastasis (between 50% and 65%) in patients with
PD-IDC.1,9,16-20 Yet, because PD-IDC characteristically
includes high-grade, hormone receptor-negative tumors
of large size with HER2/neu overexpression, it is difficult
to discern whether PD itself serves as an independent risk
factor for lymph node metastases and a poor prognosis.
Studies exploring the overall prognostic significance
of PD have suggested a negative association with breast
cancer survival. In 2002, Kothari et al performed a
matched case-control study between 40 patients with PD-
IDC and a control group of 120 women with invasive
breast cancer, comparing OS after matching for age,
grade, tumor size, and lymph node status.21 Their find-
ings included a 10-year OS rate of 49% for patients with
PD-IDC, which was significantly worse than that in the
comparison group of patients who had breast cancer with-
out PD (10-year OS rate, 64%). In another matched case-
control study that compared 32 patients who had PD
with a control group of patients who had standard breast
cancer, Ortiz-Pagan et al reported significantly lower OS
at 5 years, although it is interesting to note that this effect
did not translate into lower disease-free survival.13 More
recently, a 2013 case-control study from Ling et al com-
paring a group of 52 patients with PD versus a control
group of 156 patients reported lower 5-year relapse-free
survival (52.2% vs 81.4%; P < .01) and BCSS (62.1% vs
85.9%) rates for the patients who had PD with underlying
invasive carcinoma versus those who had invasive carci-
noma alone.12
Two previous population-based studies have exam-
ined survival in patients with PD. In 2006, Chen et al
used SEER data to report a 15-year BCSS rate of 61%
(95% CI, 53%-68%) in 859 women who had PD and
IDC diagnosed between 1988 and 2002. Those authors
also observed no significant difference in survival when
they compared women who underwent central lumpec-
tomy and mastectomy after adjusting for tumor size and
lymph node status.7 In 2008, Dalberg and colleagues
reported results from a population-based cohort in Swe-
den consisting of 68 women with PD and invasive cancer

4338 Cancer December 15, 2015

 Survival in Paget Disease of the Breast/Wong et al

TABLE 4. Cox Regression Analysis for Disease-Free 62.1%, and have similarly noted the anatomy of the main
Survival, N 5 292,899b lymphatic vessels that course through the retroareolar
Characteristic Adjusted HRs for BCSS (95% CI) region as a possible causal explanation for this. 22,23
Another potential explanation may include HER2/
Histology neu overexpression in PD-IDC tumors, which may
IDC alone 1.00
PD-IDC 0.91 (0.77-1.07) enhance the cell proliferation and intraepithelial migra-
Age at diagnosis, y tion of neoplastic cells.24,25 In their case-control study,
<40 1.00
40-59 0.89 (0.85-0.93)b
Kothari et al attributed poorer survival to increased
60-79 1.22 (1.16-1.28)b HER2/neu positivity in the PD group, and reported that
80 2.41 (2.26-2.56 b
Tumor size, cm
controlling for HER2/neu status in addition to age, tumor
<2 1.00 size, tumor grade, and lymph node status resulted in simi-
2-5 1.98 (1.92-2.05)b lar OS rates between PD patients and matched controls.
>5 3.15 (3.00-3.30)b
No. of positive lymph nodes Among the patients who had HER2/neu status available,
0 1.00 HER2/neu overexpression was present in approximately
1-3 1.96 (1.89-2.03)b
4-9 3.59 (344-3.74)b 60% of those who had PD-IDC, a value 2 to 4 times
10 5.75 (5.49-6.03)b higher than that of the patients who had IDC alone
Tumor grade
1 1.00
reported in our study and in the literature.26 Unfortu-
2 1.91 (1.78-2.06)b nately, however, because the inclusion of HER2/neu sta-
3 2.98 (2.77-3.20)b
Hormone receptor status
tus in SEER was limited to the end of our study period
ER1/PR1 1.00 (2010-2011), we were unable to further explore the rela-
ER1/PR2 1.50 (1.43-1.57)b tion of HER2/neu positivity on long-term survival out-
ER-/PR1 1.95 (1.78-2.14)b
ER2/PR2 2.21 (2.14-2.28)b comes in patients with PD. Additional insight would be
Surgical therapy gained from more extended follow-up of patients who
Breast conserving 1.00
Mastectomy 1.10 (1.06-1.14)b have HER2/neu data available, including further analysis
Radiation therapy of outcomes according to tumor biologic subtype.
No 1.00
Yes 0.79 (0.77-0.81)b
The strengths of our study include its sample size,
which represents the largest cohort to date of PD patients
Abbreviations: 2, negative; 1, positive; BCSS, breast cancer-specific sur- with underlying IDC on whom survival data have been
vival; CI, confidence interval; ER, estrogen receptor; HR, hazard ratio; IDC,
invasive ductal carcinoma; PD-IDC, Paget disease with invasive ductal car- studied. However, these findings must be interpreted within
cinoma; PR, progesterone receptor.
a
Patients with unknown survival time or unknown number of positive lymph
the context of the data. First, SEER does not have measures
nodes excluded from the analysis. in place for central pathology review, which increases the
b
After adjusting for all other variables in the table, these values were statis- possibility of misclassification bias, particularly in a rare
tically significant (P <.05).
pathologic entity such as PD. Second, SEER does not col-
and reported a 10-year BCSS rate of 75%.2 The results of lect certain pathologic characteristics, such as multifocality
our study are in agreement with these and other values or lymphovascular invasion, 2 factors known to be predic-
reported in the recent literature, with a 10-year BCSS of tive of lymph node metastasis. Thus, the observed associa-
77.9% for all patients who had stage I to III PD-IDC. tion between PD and increased lymph node positivity
In our adjusted analysis, we observed that axillary reported here may be explained through these unmeasured
metastases were 1.83 times more likely to occur in patients confounders. Third, our survival analysis was limited by a
who had PD compared those who had IDC alone. How- lack of information on systemic therapy and limited years of
ever, questions remain concerning why PD is associated follow-up for many patients. This may be particularly appa-
with increased rates of axillary metastasis or, conversely, rent for those with HER2/neu-positive disease, of whom the
why patients who have aggressive neoplasms with axillary patients with PD-IDC were represented in greater propor-
metastasis develop PD. One possible explanation for the tion, and the use of adjuvant trastuzumab would have
higher rates of lymph node involvement may include the resulted in improved outcomes.27-29 Thus, the effect of
central location of PD, with its proximity and potential HER2/neu-targeted therapy may be attenuating the differ-
invasion of the rich lymphovascular plexus that exists ence in survival reported from earlier studies.
beneath the nipple-areola complex. Previous studies have Despite the stated limitations, our study demon-
demonstrated high rates of lymph node metastases in cen- strates that, compared with women who have IDC alone,
trally located, retroareolar tumors, ranging from 32.3% to patients who have PD plus underlying IDC are more

Cancer December 15, 2015 4339

Original Article
likely to present with axillary lymph node metastases.
However, although PD appears to be associated with
more advanced disease, it does not appear to alter progno-
sis within a given stage after controlling for patient age, tu-
mor characteristics, and local therapy.
FUNDING SUPPORT
Dr. Wong was supported by a Henry R. Shibata Fellowship from
the Cedars Cancer Institute.
CONFLICT OF INTEREST DISCLOSURES
Dr. Freedman reports institutional funding for studies from Gen-
entech and Puma outside the submitted work.
REFERENCES
1. Ashikari R, Park K, Huvos AG, Urban JA. Paget’s disease of the
breast. Cancer. 1970;26:680-685.
2. Dalberg K, Hellborg H, Warnberg F. Paget’s disease of the nipple
in a population based cohort. Breast Cancer Res Treat. 2008;111:
313-319.
3. Sandoval-Leon AC, Drews-Elger K, Gomez-Fernandez CR, et al.
Paget’s disease of the nipple. Breast Cancer Res Treat. 2013;141:1-12.
4. Paone JF, Baker RR. Pathogenesis and treatment of Paget’s disease
of the breast. Cancer. 1981;48:825-829.
5. Yim JH, Wick MR, Philpott GW, et al. Underlying pathology in
mammary Paget’s disease. Ann Surg Oncol. 1997;4:287-292.
6. Lagios MD, Westdahl PR, Rose MR, Concannon S. Paget’s disease of
the nipple. Alternative management in cases without or with minimal
extent of underlying breast carcinoma. Cancer. 1984;54:545-551.
7. Chen CY, Sun LM, Anderson BO. Paget disease of the breast:
changing patterns of incidence, clinical presentation, and treatment
in the United States. Cancer. 2006;107:1448-1458.
8. Fu W, Mittel VK, Young SC. Paget disease of the breast: analysis of
41 patients. Am J Clin Oncol. 2001;24:397-400.
9. Sakorafas GH, Blanchard K, Sarr MG, Farley DR. Paget’s disease of
the breast. Cancer Treat Rev. 2001;27:9-18.
10. Sek P, Zawrocki A, Biernat W, Piekarski JH. HER2 molecular sub-
type is a dominant subtype of mammary Paget’s cells. An immuno-
histochemical study. Histopathology. 2010;57:564-571.
11. Durkan B, Bresee C, Bose S, et al. Paget’s disease of the nipple with
parenchymal ductal carcinoma in situ is associated with worse prog-
nosis than Paget’s disease alone. Am Surg. 2013;79:1009-1012.
12. Ling H, Hu X, Xu XL, et al. Patients with nipple-areola Paget’s disease
and underlying invasive breast carcinoma have very poor survival: a
matched cohort study [serial online]. PLoS One. 2013;8:e61455.
4340
13. Ortiz-Pagan S, Cunto-Amesty G, Narayan S, et al. Effect of Paget’s
disease on survival in breast cancer: an exploratory study. Arch Surg.
2011;146:1267-1270.
14. Edge S, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, eds.
AJCC Cancer Staging Manual. 7th ed. New York: Springer; 2010.
15. Wang J, Mittendorf EA, Sahin AA, et al. Outcomes of sentinel
lymph node dissection alone vs. axillary lymph node dissection in
early stage invasive lobular carcinoma: a retrospective study of the
Surveillance, Epidemiology, and End Results (SEER) database [serial
online]. PLoS One. 2014;9:e89778.
16. Caliskan M, Gatti G, Sosnovskikh I, et al. Paget’s disease of the
breast: the experience of the European Institute of Oncology and
review of the literature. Breast Cancer Res Treat. 2008;112:513-521.
17. Chaudary MA, Millis RR, Lane EB, Miller NA. Paget’s disease of
the nipple: a 10-year review including clinical, pathological, and im-
munohistochemical findings. Breast Cancer Res Treat. 1986;8:139-
146.
18. Nance FC, DeLoach DH, Welsh RA, Becker WF. Paget’s disease of
the breast. Ann Surg. 1970;171:864-874.
19. Siponen E, Hukkinen K, Heikkila P, et al. Surgical treatment in
Paget’s disease of the breast. Am J Surg. 2010;200:241-246.
20. Kollmorgen DR, Varanasi JS, Edge SB, Carson WE 3rd. Paget’s dis-
ease of the breast: a 33-year experience. J Am Coll Surg. 1998;187:
171-177.
21. Kothari AS, Beechey-Newman N, Hamed H, et al. Paget disease of
the nipple: a multifocal manifestation of higher-risk disease. Cancer.
2002;95:1-7.
22. Bevilacqua J, Cody H 3rd, MacDonald KA, et al. A prospective vali-
dated model for predicting axillary node metastases based on 2,000
sentinel node procedures: the role of tumour location [corrected].
Eur J Surg Oncol. 2002;28:490-500.
23. Yoshihara E, Smeets A, Laenen A, et al. Predictors of axillary lymph
node metastases in early breast cancer and their applicability in clini-
cal practice. Breast. 2013;22:357-361.
24. Wolber RA, Dupuis BA, Wick MR. Expression of c-erbB-2 onco-
protein in mammary and extramammary Paget’s disease. Am J Clin
Pathol. 1991;96:243-247.
25. Wolf-Yadlin A, Kumar N, Zhang Y, et al. Effects of HER2 overex-
pression on cell signaling networks governing proliferation and
migration [serial online]. Mol Syst Biol. 2006;2:54.
26. Reese DM, Slamon DJ. HER-2/neu signal transduction in human
breast and ovarian cancer. Stem Cells. 1997;15:1-8.
27. Joensuu H, Kellokumpu-Lehtinen PL, Bono P, et al. Adjuvant doce-
taxel or vinorelbine with or without trastuzumab for breast cancer.
N Engl J Med. 2006;354:809-820.
28. Piccart-Gebhart MJ, Procter M, Leyland-Jones B, et al. Trastuzumab
after adjuvant chemotherapy in HER2-positive breast cancer. N Engl
J Med. 2005;353:1659-1672.
29. Romond EH, Perez EA, Bryant J, et al. Trastuzumab plus adjuvant
chemotherapy for operable HER2-positive breast cancer. N Engl J
Med. 2005;353:1673-1684.
Cancer December 15, 2015