Paget disease of the breast: Changing patterns of incidence, clinical presentation, and treatment in the U.S.

Authors

  • Chin-Yau Chen MD,

    1. Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington
    2. Section of Breast Surgery, Department of Surgery, Keelung Hospital, Department of Health, Executive Yuan, Keelung, Taiwan
    3. Department of Surgery, Taipei Veterans General Hospital, Taipei, Taiwan
    Search for more papers by this author
  • Li-Min Sun MD, MPH,

    1. Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington
    2. Department of Epidemiology, University of Washington, Seattle, Washington
    3. Department of Radiation Oncology, Chang-Gung Memorial Hospital, Kaohsiung, Taiwan
    Search for more papers by this author
  • Benjamin O. Anderson MD

    Corresponding author
    1. Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington
    2. Department of Surgery, University of Washington, Seattle, Washington
    • Department of Surgery, Box 356410, University of Washington, Seattle, WA 98195
    Search for more papers by this author
    • Fax: (206) 543–8136.


Abstract

BACKGROUND.

Paget disease is an uncommon presentation of breast cancer that increased in incidence in the U.S. between 1973 and 1987. Characterized by malignant crusting or ulceration of the nipple, Paget disease can present in 1 of 3 ways: 1) in conjunction with an underlying invasive cancer, 2) in conjunction with underlying ductal carcinoma in situ (DCIS), or 3) alone without any underlying invasive breast carcinoma or DCIS. Paget disease can be treated with breast conservation by undergoing central lumpectomy. The objective of this study was to determine how the incidence, presentation, biomarkers, operative approach, and outcome of Paget disease have evolved in the U.S. since 1988.

METHODS.

Between 1988 and 2002, 1738 women who were diagnosed with Paget disease were reported in the 9 registries of the Surveillance, Epidemiology, and End Results Program. To the authors' knowledge, the current study on tumor characteristics, surgical intervention, and survival represented the largest series of Paget disease ever reported.

RESULTS.

Although the overall incidence of breast cancer increased between 1988 and 2002, the incidence of Paget disease concomitantly decreased by 45% (95% confidence interval, from −35% to −53%). This decreasing incidence was greatest for Paget disease associated with invasive cancer or DCIS. Invasive cancer associated with Paget disease more commonly was estrogen receptor negative, progesterone receptor negative, and of high histologic grade. Even when 60% of the disease was located centrally, only 293 of 1642 patients with Paget disease (18%) who were treated surgically underwent central lumpectomy. Patients with Paget disease who underwent breast conservation had outcomes equivalent to the outcomes among patients who underwent mastectomy.

CONCLUSIONS.

The incidence of Paget disease associated with underlying invasive cancer or DCIS decreased since 1988. Patients who underwent central lumpectomy and patients who underwent mastectomy for Paget disease had similar outcomes; nonetheless, most patients who were candidates for preservation underwent mastectomy. Cancer 2006. © 2006 American Cancer Society.

Paget disease of the breast originally was described by Paget in 1874 as nipple ulceration associated with an underlying breast carcinoma mass. The largest population-based reference resource for breast cancer in the U.S. is the Surveillance, Epidemiology, and End Results (SEER) registry of the National Cancer Institute. Of 157,546 women with invasive breast cancer who were registered in the SEER data base between 1973 and 1987, there were 1763 women (1.1%) who had histologically confirmed Paget disease.1 That early report suggested that the incidence of Paget disease has been rising in the U.S., although those findings were based on SEER data from the years 1973 to 1987.

Paget disease may occur in the nipple 1) in conjunction with an invasive cancer mass, 2) with underlying ductal carcinoma in situ (DCIS), or 3) alone without any underlying invasive breast carcinoma or DCIS. The associated underlying cancer may be located centrally in the breast adjacent to the nipple, or it may be located peripherally in the breast. Early reports described Paget disease alone without an underlying cancer as rare, representing at most 8% of patients with Paget disease.2–4 Prior SEER data-based reports provided no assessment of whether changes in the incidence were the same for women who had Paget disease with underlying invasive carcinoma, Paget disease with underlying DCIS, and Paget disease without underlying carcinoma.

In the prior analysis of SEER data from 1973 through 1984, the mean age of women with Paget disease was 62 years and did not differ significantly from the mean age of women with ductal breast cancer (61 years).5 In a retrospective study of 37 patients with Paget disease, 24 of 37 patients (65%) had high-grade, invasive disease.3 In another small series, investigators observed that invasive cancers associated with Paget disease predominantly were negative for both estrogen receptor (ER) and progesterone receptor (PR).6 Other than the findings from those studies, to our knowledge, the dominant tumor characteristics of patients with Paget disease have not been analyzed in a large, population-based study.

Although it was demonstrated in the 1980s that breast conservation had a therapeutic benefit equivalent to that of mastectomy, surgeons have been reluctant to accept breast-conserving procedures for Paget disease.7 Proponents of mastectomy cite the high frequency of occult breast malignancy identified in surgical specimens. Retrospective studies of Paget disease have identified clinically and/or mammographically occult lesions in areas far away from the nipple.7 Historically, surgeons also may have considered breast conservation less desirable for patients whose nipple and areola had to be removed. Nonetheless, in selected patients, Paget disease is amenable to breast-conserving surgery using the central lumpectomy, in which the nipple-areolar complex is excised in conjunction with the lumpectomy specimen.8 Therefore, using the SEER data base, we undertook a comparison of survival between women who underwent breast-conserving surgery (that is, central lumpectomy) and women who underwent mastectomy. We included a more complete assessment of long-term survival results, which may yield information useful for clinical decision making in the treatment of patients with this disease. To our knowledge, the current analysis represents the largest study ever reported on Paget disease of the breast.

The objective of this study was to assess how the presentation, treatment, and outcome of Paget disease of the breast has evolved in the U.S. since the middle 1980s. SEER registry data were analyzed to determine how the incidence of Paget disease of the breast changed compared with breast cancer without Paget disease and to assess whether these changes were similar for Paget disease with underlying invasive carcinoma, for Paget disease with underlying DCIS, and for Paget disease without underlying carcinoma. For Paget disease with underlying invasive cancer, the pathologic features of that invasive disease (tumor size, lymph node status, histologic grade, and hormone receptor status) were compared with the same features for invasive cancers that were not associated with Paget disease. Finally, the use and outcome of breast-conservation surgery versus mastectomy also were compared in this population-based study.

MATERIALS AND METHODS

Data Sources

We examined the publicly available records of the 9 population-based registries in the SEER 9 Registries data base (November 2004 submission).9 The registries are located in Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco-Oakland, Seattle-Puget Sound, and Utah. SEER provides information on patient demographics and tumor characteristics of newly diagnosed malignancies for approximately 10% of the U.S. population through these 9 registries. The SEER Program has been expanded up to 13 registries, but the 4 new registries can offer data only from 1992 on. For long-term evaluation of incidence trends, we chose the SEER 9 Registry as our study data base.

SEER Case Selection and Definition

Male Paget disease was reported at a much lower incidence rate than in females1, 7 and was not included in this study. We restricted our analysis to women with no prior history of any type of cancer. All tumors were confirmed microscopically as representing true Paget disease.

Histopathology data from all years of the study were merged into a common coding system for the purpose of statistical comparison. Although the SEER Program has used several different coding systems for cancer morphology over time, the histology definitions for ductal carcinoma and subgroups of Paget disease were the same as those in the International Classification of Disease for Oncology 1st edition (ICD-O), 2nd edition (ICD-O-2), and 3rd edition (ICD-O-3).9, 10 Prior to the introduction of the ICD-O in 1976, histology was coded by using the Manual of Tumor Nomenclature and Coding, in which all the subgroups of Paget disease of the breast were defined under a single category. We chose to merge data into the ICD-O-2 coding system (1992–2000) as our histology classification, because it covered most of the years in our study. The histologic classifications from ICD-O and ICD-O-3 were converted to ICD-O-2 codes by using the SEER conversion program.11, 12 Four ICD-O-2 histology codes were used in this study: code 8500 (ductal carcinoma), code 8540 (mammary Paget disease), code 8541 (Paget disease with infiltrating ductal carcinoma), and code 8543 (Paget disease with intraductal carcinoma). Extramammary Paget disease (code 8542) and Paget disease of bone were not included. Paget disease of the nipple with the underlying tumor other than ductal carcinoma was coded according to that specific underlying tumor and could not be searched. For standard ductal carcinoma, behavior codes (malignant [3] or in situ [2]) were used to separate invasive ductal carcinoma from DCIS.

We identified 1738 women who had Paget disease of the breast diagnosed from January 1, 1988 to December 31, 2002 for the analysis of overall incidence trend. In total, 34 patients who had Paget disease alone or Paget disease with DCIS reportedly had positive pathologic lymph nodes. Those 34 patients had disease that was regarded as questionable biologically; therefore they were excluded from the analysis. Overall, there were 1704 eligible patients for analysis including 859 patients (50.4%) who had Paget disease with infiltrating ductal carcinoma, 618 patients (36.3%) who had Paget disease with DCIS, and 227 patients (13.3%) who had Paget disease alone. Because it is believed that Paget disease is derived from an underlying ductal carcinoma, we enrolled all women with standard invasive ductal carcinoma (n = 155,965 women) and standard DCIS (n = 21,426 women) who were diagnosed during the same 15-year period for control group comparisons.

Study Variables

Patient demographics included age at diagnosis (age <60 years vs. age ≥60 years) and race (white vs. nonwhite). The median age at diagnosis for patients with Paget disease was 64 years. The sites of primary tumor location were defined as “central” for both code C50.0 (nipple) and code C50.1 (central or subareolar portion of breast). The primary tumor location was defined as “peripheral” for code C50.2 (upper inner quadrant), code C50.3 (lower inner quadrant), code C50.4 (upper outer quadrant), code C50.5 (lower outer quadrant), code C50.6 (axillary tail), and code C50.8 (overlapping lesion of breast). Tumor characteristics included tumor size (≤2 cm, 2–5 cm, or >5 cm) for patients who had Paget disease with underlying tumor, regional lymph node status (negative vs. positive in pathologic report, regardless of its level and number), ER and PR expression (positive vs. negative), and tumor grade. The SEER 12-digit Extent of Disease codes were used for analyses of tumor size and lymph node status. Tumor grade was defined according to ICD-O-2 coding conversions.10 For patients who were diagnosed after 1996, the (Scarff) Bloom–Richardson (BR) grading system was approved as a coding reference in SEER. The BR low grade, intermediate grade, and high grade were coded to 1 (well differentiated), 2 (moderately differentiated), and 3 (poorly differentiated), respectively.13 In this study, grade code 4 (undifferentiated; anaplastic) and code 3 were defined as high grade, and grade codes 1 and 2 were defined as low grade. Data for tumor markers of ER and PR were available only after 1990.

In total, 1642 women with Paget disease reportedly underwent surgery. Women were classified as having undergone central lumpectomy if, according to the SEER registry, they underwent partial (less than total) mastectomy (which includes segmental mastectomy, lumpectomy, quadrantectomy, tylectomy, wedge resection, nipple resection, excisional biopsy, and partial mastectomy). Mastectomy included subcutaneous mastectomy, simple mastectomy, modified radical mastectomy, radical mastectomy, and extended radical mastectomy.

Statistical Analysis

The National Cancer Institute's SEER*Stat software package (version 6.1.4; built on April 13, 2005) was used to calculate incidence rates.14 The incidence rates were expressed per 100,000 woman-years and were age adjusted according to the 2000 U.S. standard population (19 age groups; Census P25-1130).9 The proportional changes of incidence rates were calculated by using negative binomial regression.15 We estimated linear trends in incidence rates by subgroups of Paget disease over the 15 years from 1988 to 2002. All analyses were adjusted for age at diagnosis according to 19 SEER age groups (ages birth, 1–4 years, 5–9 years, 10–14 years, 15–19 years, 20–24 years, 25–29 years, 30–34 years, 35–39 years, 40–44 years, 45–49 years, 50–54 years, 55–59 years, 60–64 years, 65–69 years, 70–74 years, 75–79 years, 80–84 years, and 85 years or older). Low-risk prognostic factors were assigned as the reference at a rate ratio of 1.0; then, rate ratios were calculated by dividing the age-adjusted breast cancer incidence rate by the incidence rate in the reference.

The SEER 9 Registries data base was accessed in ASCII format and was analyzed with the Stata statistical software package (version 9.1 for Windows; StataCorp., College Station, TX). Differences in the mean age at diagnosis for breast cancer subtypes were evaluated by using the Student t test and a 1-way analysis of variance. We conducted the analyses using breast carcinoma-specific mortality. Survival was calculated in months from the date of diagnosis to whichever of the following occurred first: the date of death, the date last known to be alive, or December 31, 2002, which was the follow-up cut-off date used in our analysis. We estimated survival curves by using the Kaplan–Meier product-limit method.16 The log-rank test was used to assess the association of survival with different subgroups and different surgical types. Cox regression analysis was performed to compute hazard ratios and 95% confidence intervals (95% CIs) and to evaluate the effects of confounding factors. The Cox proportional hazards model assumption was tested by using scaled Schoenfeld residuals.17 In all statistical tests, P values <.05 were considered statistically significant. All P values were 2-sided.

RESULTS

The age-adjusted incidence rates of female Paget disease of the breast peaked in 1985 and decreased yearly thereafter (Fig. 1). The age-adjusted incidence rates decreased from 1.31 to 0.64 per 100,000 woman-years between 1988 and 2002 (Table 1). The entire decline of age-adjusted incidence rates was 45% (95% CI, from −35% to −53%; P < .001). By comparison, the incidence rates of ductal carcinoma, including invasive cancer and DCIS, increased by 10% (95% CI, 7–14%; P<.001) during the same period (1988–2002; data not shown).

Figure 1.

Age-adjusted incidence rates are illustrated for women with Paget disease of the breast (per 100,000 woman-years; data from the Surveillance, Epidemiology, and End Results 9 Registries data base, 1973– 2002). From 1988 to 2002, incidence rates decreased by 45% (95% confidence interval, from −35% to −53%; P < .001; binomial negative regression).

Table 1. Age-Adjusted Incidence Rates of Paget Disease of the Breast: Surveillance, Epidemiology, and End Results 9 Registries Data Base*
YearTotal population in registriesNo. of patientsAge-adjusted incidence rate per 100,00095% CI
  • 95% CI indicates 95% confidence interval.

  • *

    Rates were age-adjusted to the 2000 U.S. standard population and are expressed per 100,000 woman-years.

  • Proportional changes were estimated by using binomial negative regression and were adjusted for 19 age groups.

198811,804,5791471.311.10–1.55
198911,925,1331351.170.98–1.39
199012,074,6561351.160.97–1.38
199112,243,8591150.960.79–1.15
199212,422,5761150.920.76–1.11
199312,591,9351080.860.70–1.04
199412,743,1771220.960.80–1.15
199512,896,9531100.860.70–1.03
199613,051,3221180.900.74–1.08
199713,207,0821130.860.71–1.03
199813,350,7041000.740.61–0.91
199913,471,0981140.820.68–0.99
200013,597,107970.690.56–0.85
200113,715,1301160.820.67–0.98
200213,798,019930.640.52–0.79
Proportional change  −0.45−0.35 to −0.53

Both histologic subgroups in which Paget disease was associated with underlying cancer (Paget disease with underlying invasive ductal carcinoma and Paget disease with underlying DCIS) decreased in incidence between 1988 and 2002 (Fig. 2). Their age-adjusted incidence rates decreased by 49% (95% CI, from −35% to −59%; P < .001) and 44% (95% CI, from −27% to −58%; P < .001), respectively. By comparison, the incidence change in Paget disease without any underlying invasive carcinoma or DCIS (Paget disease alone) was not significantly different from 0 (proportional change, −21%; 95% CI, from −50% to + 25%; P = .312), suggesting a stable incidence rate in the 15 years between 1988 and 2002. Consequently, the relative proportion of Paget disease without underlying carcinoma increased from 12% in 1988 to 15% in 2002.

Figure 2.

Age-adjusted incidence rates are illustrated for women with Paget disease of breast (per 100,000 woman-years) stratified by Paget disease with infiltrating ductal carcinoma, Paget disease with ductal carcinoma in situ, and Paget disease alone. P values were calculated by using binomial negative regression (data from the Surveillance, Epidemiology, and End Results 9 Registries data base, 1988–2002).

The mean age at diagnosis for women with Paget disease was 62.6 years (95% CI, 61.9–63.3 years). Patients who presented with Paget disease alone were significantly older than patients who presented with Paget disease associated with underlying carcinoma. Specifically, the mean ages were 60.8 years (95% CI, 59.7–61.8 years) for women who had Paget disease with infiltrating ductal carcinoma, 63.8 years (95% CI, 62.6–64.9 years) for women who had Paget disease with DCIS, and 66.2 years (95% CI, 64.4–68.0 years) for women who had Paget disease without underlying carcinoma.

The demographics and tumor characteristics of patients with Paget disease who had underlying invasive carcinoma were compared with those from patients who had standard infiltrating ductal carcinoma without Paget disease (Table 2). Histologically high-grade tumors were more observed commonly among women who had Paget disease with infiltrating ductal carcinoma (rate ratio, 1.7; 95% CI, 1.5–2.1) compared with women who had standard infiltrating ductal carcinoma (rate ratio, 0.744; 95% CI, 0.735–0.752). Compared with women who had standard infiltrating ductal carcinoma, women who had Paget disease with infiltrating ductal carcinoma were more likely to present with larger tumors, positive lymph nodes (rate ratio, 0.8; 95% CI, 0.7–0.9), negative ER status (rate ratio, 1.1; 95% CI, 0.9–1.3), and negative PR status (rate ratio, 1.4; 95% CI, 1.2–1.7). A significantly smaller fraction of women who had Paget disease with underlying DCIS presented with peripherally located disease (rate ratio for peripheral location vs. central location, 0.31; 95% CI, 0.25–0.39) compared with women who had DCIS without Paget disease (rate ratio, 2.9; 95% CI, 2.8–3.1) (Table 3).

Table 2. Breast Cancer Incidence Rates by Patient Demographics, Tumor Characteristics, and Hormone Receptor Status in Women with Infiltrating Ductal Carcinoma and in Women with Paget Disease plus Infiltrating Ductal Carcinoma: Surveillance, Epidemiology, and End Results 9 Registries Data Base, 1988–2002
CharacteristicInfiltrating ductal carcinoma (n = 155,965 Women)Paget disease plus infiltrating ductal carcinoma (n = 859 Women)
No. of patientsRate*RR (95% CI)No. of patientsRate*RR (95% CI)
  • RR indicates rate ratio; 95% CI, 95% confidence interval; ER, estrogen receptor; PR, progesterone receptor.

  • *

    Rates refer to age-adjusted incidence rates according to the 2000 U.S. standard population and are expressed per 100,000 woman-years.

  • Statistics could not be calculated.

  • c

    §Data were available only after 1990.

Age group, y
 <6074,35649.71.0 (Referent)3960.31.0 (Referent)
 ≥6081,609243.44.90 (4.85–4.95)4631.45.3 (4.6–6.0)
Race
 White131,59884.01.0 (Referent)7180.51.0 (Referent)
 Nonwhite23,75668.10.81 (0.80–0.82)1370.40.9 (0.7–1.0)
 Unknown611 4 
Tumor grade
 Low grade73,38338.41.0 (Referent)2220.11.0 (Referent)
 High grade54,20228.50.744 (0.735–0.752)3790.21.7 (1.5–2.1)
 Unknown28,38014.8 2580.1 
Tumor size, cm
 ≤294,64249.61.0 (Referent)4070.21.0 (Referent)
 2–543,19822.60.455 (0.450–0.460)2440.10.6 (0.5–0.7)
 >580714.20.085 (0.083–0.087)990.10.25 (0.20–0.31)
 Unknown10,0545.3 1090.1 
Pathologic lymph node status
 Negative84,99144.81.0 (Referent)4040.21.0 (Referent)
 Positive43,99923.30.52 (0.51–0.53)3150.20.8 (0.7–0.9)
 Unknown26,97513.6 1400.1 
ER status§
 Positive87,23645.71.0 (Referent)2710.11.0 (Referent)
 Negative28,33915.00.330 (0.325–0.334)2910.21.1 (0.9–1.3)
 Unknown40,39021.0 2970.2 
PR status§
 Positive74,94439.31.0 (Referent)2310.11.0 (Referent)
 Negative38,23420.10.512 (0.506–0.519)3300.21.4 (1.2–1.7)
 Unknown42,78722.2 2980.2 
Primary tumor location
 Central10,5885.51.0 (Referent)2320.11.0 (Referent)
 Peripheral121,56963.811.6 (11.3–1.8)4060.21.8 (1.5–2.1)
 Unknown23,80812.4 2210.1 
Table 3. Breast Cancer Incidence Rates by Patient Demographics and Primary Tumor Location in Women with Ductal Carcinoma in Situ and in Women with Paget Disease plus Ductal Carcinoma in Situ: Surveillance, Epidemiology, and End Results 9 Registries Data Base, 1988–2002
CharacteristicDCIS (n = 21,426 Women)Paget disease plus DCIS (n = 618 Women)
No. of patientsRate*RR95% CINo. of patientsRate*RR95% CI
  • DCIS indicates ductal carcinoma in situ; RR, rate ratio; 95% CI, 95% confidence interval.

  • *

    Rates refer to age-adjusted incidence rates according to the 2000 U.S. standard population and are expressed per 100,000 woman-years.

  • Statistics could not be calculated.

Age group, y
 <6011,7127.91.0Referent2270.21.0Referent
 ≥60971429.23.73.6–3.83911.27.76.5–9.1
Race
 White17,38711.41.0Referent5360.31.0Referent
 Nonwhite385711.20.980.95–1.02780.2 0.5–0.9
 Unknown182  4  
Primary tumor location
 Central16580.91.0Referent3860.21.0Referent
 Peripheral14,9408.02.92.8–3.11120.10.310.25–0.39
 Unknown48282.6  1200.1  

Kaplan–Meier plots (Fig. 3) showed that the 15-year breast cancer-specific survival rates were 61% (95% CI, 53–68%) for women who had Paget disease with infiltrating ductal carcinoma, 94% (95% CI, 89–97%) for women who had Paget disease with DCIS, and 88% (95% CI, 77–94%) for women who had Paget disease without underlying carcinoma. Because the data for tumor grade, tumor size, ER status, and PR status were available only for the women who had Paget disease with infiltrating ductal carcinoma, univariate and multivariate analyses (Table 4) of prognostic indicators for survival were undertaken in this subgroup of patients. The proportional hazards assumption was verified. Although tumor grade, tumor size, lymph node status, and type of surgery (central lumpectomy vs. mastectomy) all were significant factors for survival in univariate analyses, multivariate analyses revealed that only tumor size and lymph node status were statistically significant, independent prognostic indicators of disease-specific breast cancer mortality.

Figure 3.

These Kaplan–Meier plots illustrate 15-year breast cancer-specific survival curves for women who had Paget disease with ductal carcinoma in situ, Paget disease alone, and Paget disease with infiltrating ductal carcinoma (data from the Surveillance, Epidemiology, and End Results 9 Registries data base, 1988–2002).

Table 4. Multivariate Cox Model for Breast Cancer Death in Women with Paget Disease plus Infiltrating Ductal Carcinoma
ParameterBreast cancer death
Risk ratio95% CIP
  1. 95% CI indicates 95% confidence interval; ER, estrogen receptor; PR, progesterone receptor.

Age, y  .294
 <601.0Referent 
 ≥601.30.8–2.1 
Race  .942
 White1.0Referent 
 Nonwhite1.00.6–1.8 
Grade  .118
 Low grade1.0Referent 
 High grade1.60.9–2.8 
Tumor size, cm
 ≤21.0Referent 
 2–52.11.1–3.9.023
 >52.91.4–6.1.004
Lymph node status  <.001
 Negative1.0Referent 
 Positive4.22.1–8.3 
ER status  .069
 Positive1.0Referent 
 Negative1.81.0–3.4 
PR status  .755
 Positive1.0Referent 
 Negative1.10.6–2.1 
Surgery type  .781
 Central lumpectomy1.0Referent 
 Mastectomy1.30.2–9.9 

Of 1642 women with Paget disease who underwent surgery, 293 patients (18%) underwent central lumpectomy, and 1349 women (82%) underwent mastectomy. Of 1232 women who underwent surgery and who had tumor location (central vs. peripheral) documented, 735 women (60%) had centrally localized disease. Of 189 women within this subgroup with centrally located tumors who had Paget disease with underlying invasive carcinoma, 118 women (62%) had small tumors (T1; ≤2 cm). Despite the central location and small invasive tumor size in this subgroup, 84% (99 of 118 women) underwent mastectomy rather than central lumpectomy (Table 5). This was a significantly higher mastectomy rate than similar patients who had centrally localized, small, invasive ductal carcinoma diagnosed and treated during the same period (52%; P < .001). Similarly, of 546 women who had Paget disease with centrally localized underlying DCIS or Paget disease alone, 64% underwent mastectomy, compared with the 37% mastectomy rate for women who had centrally localized DCIS without Paget disease during the same period (P < .001). Ninety-four percent of patients (468 of 497 women) who had Paget disease with peripherally located underlying tumors underwent mastectomy, as expected, because this represented multicentric disease within the breast.

Table 5. The Use of Central Lumpectomy versus Mastectomy for Centrally Located Tumors: A Comparison of Women Diagnosed with Ductal Carcinoma and with Paget Disease, Surveillance, Epidemiology, and End Results 9 Registries Data Base, 1988–2002
Group AnalyzedLumpectomyMastectomy
No. of patients%No. of patients%P*
  • DCIS indicates ductal carcinoma in situ.

  • *

    Pearson chi-square test.

Centrally located cancers
 Centrally located, invasive cancers ≤2 cm
  Infiltrating ductal carcinoma ≤2 cm279548307752<.001
  Paget disease with underlying infiltrating ductal carcinoma ≤2 cm19169984 
 Centrally located, invasive cancers >2 cm
  Infiltrating ductal carcinoma >2 cm78220311580.015
  Paget disease with underlying infiltrating ductal carcinoma > 2 cm686592 
 Centrally located, noninvasive cancers
  DCIS10316359537<.001
  Paget disease with underlying DCIS or without underlying disease1963635064 
 All centrally located cancers
  Ductal carcinoma (DCIS and infiltrating)460840678760<.001
  Paget disease (all groups)2213051470 
All peripherally located cancers
 Ductal carcinoma (DCIS and infiltrating)69,1695459,44546<.001
 Paget disease (all groups)29646894 

The 15-year breast cancer-specific survival rates (Fig. 4) were 92% (95% CI, 84–96%) for central lumpectomy and 94% (95% CI, 88–96%) for mastectomy for women who had Paget disease alone and women who had Paget disease with DCIS (P value not significant). The 15-year breast cancer-specific survival rates for women who had Paget disease with underlying infiltrating ductal carcinoma (Fig. 5) were 87% (95% CI, 64–96%) for central lumpectomy and 60% (95% CI, 52–67%) for mastectomy. Although the overall survival rate for central lumpectomy was better than that for mastectomy among women who had Paget disease with underlying invasive carcinoma, there was no statistical difference in survival between women who underwent central lumpectomy and women who underwent mastectomy after adjusting for tumor size and lymph node status.

Figure 4.

These Kaplan–Meier plots illustrate 15-year breast cancer-specific survival curves for central lumpectomy versus mastectomy in women who had Paget disease alone and women who had Paget disease with ductal carcinoma in situ of the breast (data from the Surveillance, Epidemiology, and End Results 9 Registries data base, 1988–2002).

Figure 5.

These Kaplan–Meier plots illustrate 15-year breast cancer-specific survival curves for central lumpectomy versus mastectomy in women who had Paget disease with infiltrating ductal carcinoma of the breast (data from the Surveillance, Epidemiology, and End Results 9 Registries data base, 1988–2002).

DISCUSSION

In his original 1874 report, Sir James Paget described the eczematous lesions of the nipple associated with an underlying breast carcinoma.7, 18 In 1881, Thin observed that the nipple lesion contained malignant cells that were related to the underlying cancer,19 suggesting the process of intraductal extension of cancer through the major lactiferous sinuses that we know today as “Pagetoid spread.” Histologically, Paget cells are large cells with pale, clear cytoplasm and enlarged nucleoli located within the epidermis and along the basal layer. The most widely accepted hypothesis regarding the origin of Paget cells is the epidermatropic theory, which maintains that Paget cells are derived from an underlying mammary adenocarcinoma.7, 20 Evidence supporting the epidermatropic theory is based on studies showing that Paget disease is associated with an underlying breast carcinoma in most patients.2, 4, 21, 22 Binding of heregulin-α to its receptor on Paget cells can induce chemotaxis of these breast cancer cells, which eventually migrate into the overlying nipple epidermis.23 It is noteworthy that Paget cells and the underlying associated ductal carcinoma share the same immunohistochemical profile24 and the same patterns of gene expression.7

A competing theory, the intraepidermal transformation theory, maintains that Paget cells arise in situ from multipotential cells in the epidermal basal layer7 or from the terminal portion of the lactiferous duct at its junction with the epidermis.25, 26 Support for this theory stems from the observation that Paget disease can be diagnosed without any demonstrable underlying carcinoma.27 Furthermore, ultrastructural studies have demonstrated desmosomal attachments between Paget cells and adjacent keratinocytes, suggesting a true in situ development of Paget cells in these patients.7, 28

Female breast cancer incidence rates have increased since the 1980s.29 This increased incidence coincided with an increased use of mammography in asymptomatic women in the 1980s.30 These descriptive data suggested that early detection has played a role in the recent increase in the incidence of breast cancer in women.31 Prior studies of SEER registry data before 1987 showed that the incidence of Paget disease increased or even doubled during that time.1, 5 By contrast, our current study showed that the incidence of Paget disease actually decreased between 1988 and 2002. This decreasing incidence of Paget disease, particularly the decrease associated with underlying invasive cancer or DCIS, may have been because of the earlier detection of these lesions at a point in their evolution prior to the development of Pagetoid changes. Thus, these population-based data are consistent with the epidermatropic theory, which maintains that Paget disease develops as an extension of an underlying invasive or noninvasive cancer.

Prior reports suggested that Paget disease was associated with an underlying invasive ductal carcinoma or DCIS in from 92% to 100% of patients.2, 4, 21, 22 Although the incidence rates of Paget disease with an underlying tumor decreased steadily between 1988 and 2002, at the same time that mammographic detection of early cancers was improving, the incidence rates of Paget disease without an underlying tumor remained unchanged. Specifically, 227 of 1704 women (13%) with Paget disease had no underlying breast carcinoma. Thus, although our data are consistent with the epidermatropic theory of Paget disease etiology, our data do not disprove the intraepidermal transformation theory, which maintains that Paget disease develops in situ in the major lactiferous sinuses. Collectively, the findings of the current study support the combined hypothesis that Paget disease may develop by either epidermatropic or intraepidermal transformation in different settings.

Patients with Paget disease reportedly have poor tumor characteristics when their disease is associated with underlying invasive cancer.32 In 1 report, 24 of 37 women (65%) who had Paget disease with underlying invasive cancer had tumors with Grade 3 histology.3 In another report of 20 patients with who had mammary Paget disease, 80% of tumors showed strong overexpression (3+) of HER-2 and were negative for ER.24 In our study, patients who had invasive cancers with underlying Paget disease were more likely to have histologically high-grade tumors (rate ratio, 1.7 vs. 0.744), positive lymph node status (rate ratio, 0.8 vs. 0.52), negative ER status (rate ratio, 1.1 vs. 0.33), and negative PR status (rate ratio, 1.4 vs. 0.512) (Table 2). Thus, our results confirmed prior small studies and provided a large, population-based reference that demonstrates the relative histologic aggressiveness of invasive carcinoma associated with Paget disease.

Although women with Paget disease presented with poor tumor characteristics other those associated with than standard ductal carcinoma, our multivariate analyses showed that the size of underlying tumor and lymph node status were the only independent prognostic factors for survival. Mastectomy was not associated with better survival outcomes compared with central lumpectomy. These finding are consistent with prior studies showing that the prognosis for women with Paget disease is determined primarily by the underlying carcinoma in the breast rather than the Paget disease itself.

Kollmorgen et al. reviewed 68 patients with Paget disease of the breast who presented over 33 years.33 Those investigators defined “centrally located” underlying tumors as tumors located within 2 cm of the areolar margin. The authors reported that 56 patients (82%) had some underlying tumor, and 61% of those tumors were located centrally in the breast. In our study, of 1232 patients who underwent surgery and had their tumor location documented (central vs. peripheral), 735 patients (60%) had centrally localized disease. Of the women who had small, centrally located, underlying, invasive cancers, 84% (99 of 118 women) underwent mastectomy rather than breast-conserving central lumpectomy. This mastectomy rate was significantly higher than that for similar patients who had ductal carcinoma without Paget disease (P < .001), suggesting that surgeons specifically recommended total mastectomy rather than central lumpectomy for patients who presented with Paget disease of the nipple.

In 1987, the European Organization for Research and Treatment of Cancer initiated a prospective study of breast-conserving therapy that consisted of cone excision (central lumpectomy) followed by whole-breast radiotherapy for Paget disease without associated invasive carcinoma and observed no difference in outcome between the 2 groups.8 In another study, among 92 of 104 patients (88.5%) who underwent mastectomy and 12 of 104 patients (11.5%) who underwent a breast-conserving procedure, there were no significant differences in overall, disease-specific, or recurrence-free survival according to the type of surgery.34 In the current study, the 15-year breast cancer-specific survival rate was 92% (95% CI, 84–96%) for patients who underwent central lumpectomy and 94% (95% CI, 88–96%) for patients who underwent mastectomy for noninvasive Paget disease. The 15-years breast cancer-specific survival rate was 87% (95% CI, 64–96%) for patients who underwent central lumpectomy and 60% (95% CI, 52–67%) for patients who underwent mastectomy among those who had Paget disease with an underlying invasive carcinoma. There was no worsened survival among patients who underwent central lumpectomy rather than mastectomy. If anything, the trend for improved survival was among those patients who underwent the smaller operation. Presumably, this trend results from a selection bias in which patients with smaller cancers that portend a better prognosis were more likely to be offered breast conservation by their surgeon.

In summary, Paget disease with underlying cancer has decreased in incidence in the U.S. since 1988, whereas the incidence of Paget disease without underlying cancer has remained relatively stable. The tumor characteristics of invasive cancer associated with Paget disease are more aggressive than those of invasive cancer not associated with Paget disease, although the incidence of this subgroup has been decreasing to the greatest degree. Among 118 women who presented with invasive Paget disease that was centrally located and measured <2 cm in greatest dimension, 84% underwent mastectomy. Of the 546 women who had either Paget disease with centrally localized, underlying DCIS or Paget disease alone, 64% underwent mastectomy. Patients who underwent central lumpectomy had the same long-term survival as patients who underwent mastectomy. Breast conservation, although it requires removal of the nipple-areolar complex, may be underutilized in the U.S. and needs further investigation.

Ancillary