According to the TNM Supplement, only breast carcinoma with macroscopic ‘classic’ skin changes (e.g., edema, peau d'orange, ulceration) should be placed in the T4 category; the classification of tumors with histologically proven skin involvement but no clinical skin changes or only discreet changes should be based on the size of the tumor (T1–3). To the authors' knowledge, no data supporting these recommendations have been reported to date.
Seventy-six patients with noninflammatory breast carcinoma and histologically proven skin involvement were classified based on the degree of their skin involvement. Fifty patients (66%) presented with clinically obvious skin involvement (Group A), and 26 patients (34%) had breast carcinoma with histologically proven skin involvement but without clinical skin changes (Group B). Reclassification was undertaken to assess the malignant potential of tumors independent of the morphologic parameter ‘skin involvement’.
Patients in Group A presented significantly more frequently with extensive disease at regional or distant sites (Stages IIIC and IV; P = 0.009). The clinical outcome of patients in Group B was significantly more favorable (P = 0.0003). The adjusted 3-year survival rates in Group A and Group B were 46.7% and 92.3%, respectively, and the 5-year rates were 38.1% and 83.7%, respectively. Patients in Group B were found to have significantly better (P = 0.036) distant recurrence–free survival (DRFS) rates. The DRFS rate at 3 years was 63.6% in Group A, compared with 91.7% in Group B, and the DRFS rate at 5 years was 56.9% in Group A, compared with 82.0% in Group B. Using a cutoff point of tumor size = 3 cm, similar findings were made.
Since the introduction of the first breast carcinoma staging systems, at the beginning of the last century, cases with extension of tumor to the skin generally have been classified as locally advanced breast carcinoma (LABC), which usually is associated with poor prognosis.1 LABC accounts for approximately 10% of newly discovered breast carcinoma cases in industrialized nations and for a much greater proportion (40–60%) in developing countries.2, 3
In the current edition of the International Union Against Cancer (UICC) TNM staging system,4 which is in full agreement with the American Joint Committee on Cancer (AJCC) system,5 primary breast carcinomas with direct extension to the skin are classified as T4b lesions if they invade the skin, T4c lesions if they invade the chest wall and skin, and T4d lesions if they are inflammatory. Inflammatory carcinoma is characterized by diffuse, brawny induration of the skin with an erysipeloid edge, usually with no underlying mass.4, 5 Patients with T4 tumors of any type and without metastases (M0), regardless of axillary lymph node status, are classified as having Stage IIIB disease. According to the TNM system (Table 1), all breast carcinomas with skin involvement are included in Stage III, which is considered to be synonymous with LABC. These malignancies encompass a heterogeneous group of patients with a wide range of prognoses and clinical entities, including tumors > 5 cm with axillary lymph node metastases (T3N1–3), skin/chest wall–invading tumors (T4a–c), inflammatory breast carcinoma (T4d), and extensive regional lymph node disease (N2–3).
Table 1. TNM Staging System for Breast Carcinoma
N0, N1, N2
According to the TNM classification, only tumors accompanied by macroscopic and typically readily discernible skin changes, such as localized inflammation, edema, peau d'orange, satellite skin nodules, and ulceration, should be placed in the T4b category. It is stated explicitly that discreet skin changes, such as dimpling, retraction of the nipple, and other changes, often caused by the shortening of Cooper ligaments due to infiltration by malignant disease, may occur in T1, T2, or T3 disease and do not affect classification.4, 5 These discreet skin changes do not appear to have prognostic value.
The TNM system does not precisely define the classification of tumors that have histologically proven skin involvement but no accompanying clinical skin changes or only discreet changes.4, 5 Until recently, this subgroup of cases lay in a gray area of the TNM classification. Insufficiently detailed definitions and rules led to some leeway in the categorization of lesions, such that application of the T4b classification was not consistent. It was not until the publication of the second edition of the TNM Supplement, in 2001, that more precise recommendations for uniform use were drafted.6 These recommendations are as follows: 1) for a lesion to be classified as T4b, it is required that the previously mentioned macroscopic (clinical) features be present; and 2) microscopic invasion of the dermis alone is not sufficient for placing a lesion in the T4b category; when only microscopic invasion is present, the T classification is based solely on tumor size (T1–3).
These recommendations are based on the clinical observation that cases with only histologically verified skin involvement do not correspond, in terms of morphologic extent, clinical course, or malignant potential, to ‘classic’ cases of LABC. To the best of our knowledge, no data exist to support these recommendations, which are not yet formally included in the TNM classification system.
The goal of the current study was to determine the prognostic relevance of skin involvement in noninflammatory breast carcinoma. In the current report, we describe the prevalence and clinical course of patients who exhibit histologically proven skin involvement with and without classic clinical signs.
MATERIALS AND METHODS
Between January 1985 and March 1998, 1057 women with newly diagnosed breast carcinoma were evaluated and treated at University Women's Hospital (Basel, Switzerland). One hundred three presented with clinical and/or pathologic skin involvement (T4b/T4d) and had no history of contralateral breast carcinoma or local recurrence. Twenty-seven of these 103 patients had inflammatory breast carcinoma (T4d), which was defined by the simultaneous presence of diffuse edema, erythema, warmth, and tenderness. Tumors with localized inflammation were designated as T4b lesions.
The data on the 76 patients (7.2%) who presented with noninflammatory breast carcinoma and histologically proven skin involvement are the basis of the current analysis. The median age of these patients was 71 years (range, 40–93 years), and 78% were postmenopausal.
At the date of diagnosis, all patients were classified as having T4b disease. The median follow-up time from the date of diagnosis was 4.3 years (range, 0.1–16.8 years). All patients were followed until death or, if they remained alive, for at least 5 years.
The diagnosis of invasive carcinoma was confirmed by excisional or needle core biopsy in all cases. Hormone receptor assays and assessment of histopathologic grading according to the Nottingham modification of the Scarff–Bloom–Richardson grading scheme also were performed. Additional biologic predictive factors (e.g., overexpression of HER-2/neu, p53, and epidermal growth factor receptor), which eventually may prove to be more reliable measures of tumor aggressiveness, were assessed regularly beginning in the 1990s. Therefore, information on these factors was not available for all patients and was not considered in the final analysis. Tumor size was measured by macroscopic examination of specimens. Eight patients did not undergo surgery; for these patients, tumor size was assessed by imaging or physical examination.
Each patient underwent a staging workup, which included a recording of clinical history, physical examination, routine blood studies, chest X-ray, sonography of the liver, nuclear bone scans, and additional diagnostic studies as needed to rule out the possibility of metastatic disease. There was no standard therapeutic approach during the study period (Table 2). Treatment was selected at the discretion of the physician in charge. The spectrum of possible treatments included multidisciplinary approaches involving combined modality therapy, as well as less extensive locoregional and/or systemic therapy (e.g., tumor excision/simple mastectomy without axillary lymph node dissection, tamoxifen alone as primary treatment, etc.) for elderly patients with coexisting illnesses. In some cases, comorbid conditions limited systemic therapy or patients refused the recommended treatment. Some patients already had extensive metastatic disease, and their treatment thus was focused on maintaining quality of life while controlling symptoms.
Table 2. Treatment Data
No. of patients
Lumpectomy + axillary dissection
Mastectomy + axillary dissection
Adjuvant hormonal therapy
Neoadjuvant hormonal therapy
No systemic therapy
As a first step, clinical or pathologic staging was performed for all patients in accordance with the current (sixth) edition of the UICC/AJCC TNM classification.4, 5 Some cases had to be reclassified retrospectively, because at the time of presentation, classification rules that were different from the current ones (according to the third, fourth, or fifth edition of the TNM classification) were in effect. Based on the degree of skin involvement, patients were placed into one of two groups: 1) those with clinically obvious skin involvement, such as localized inflammation, edema, peau d'orange, and ulceration (cT4b); and 2) those with histologically proven skin involvement but no accompanying clinical changes or only discreet changes (retraction or dimpling) to the overlying skin or the nipple (cT1–3).
As a second step, reclassification was undertaken to assess the malignant potential of tumors independent of the morphologic parameter ‘skin involvement’. This clinical/histopathologic feature was no longer taken into consideration and instead was eliminated from our classification of tumors. All tumors were reclassified based on size; the category T4b was replaced with the categories T1–3. In this manner, patients who had Stage IIIB disease underwent restaging. The number of patients with Stage IIIC and Stage IV disease did not change, because the presence of extensive regional lymph node disease (N3) and distant metastases (M1) always led to categorization in Stages IIIC and IV, respectively, regardless of T classification (Table 1).
Using the Kaplan–Meier method, disease-specific survival (DSS) and distant recurrence–free survival (DRFS) were calculated from the date of diagnosis to the date of death or distant failure or, for patients who remained alive, to the date of last follow-up. Non-malignancy-related deaths were censored in the statistical analyses according to the same method used for patients who were alive and disease free. Statistical differences between groups in terms of survival curves were analyzed using the log-rank test. Comparisons between nominal parameters were made with the Fisher exact test. Continuous data were logarithmically transformed and compared using the Student t test. Statistical analyses were performed with SPSS 10.0.5 software (SPSS Inc., Chicago, IL).
Seventy-six patients with breast carcinoma and histologically proven skin involvement were included in the current review (Table 3). Fifty of these patients (66%) presented with clinically obvious skin involvement, fulfilling the criteria for the T4b classification (Group A). Twenty-six patients (34%), accounting for 2.5% of all newly diagnosed breast carcinomas within the study period, had disease with histologically proven skin involvement but no clinical skin changes or only discreet changes (Group B). The median patient age was greater in Group B than in Group A (74.5 years vs. 68.5 years).
Table 3. Patient Characteristics
Total no. of patients (%)
Premenopausal status (%)
Follow-up time (mos)
Tumor characteristics are listed in Table 4. The mean tumor diameter in Group A was 6.8 cm (range, 1.5–15.0 cm), compared with 3.0 cm (range, 1.1–7.0 cm) in Group B; the difference in tumor size was significant (P < 0.001). Hormone receptor assays were performed for all patients; 74% of patients in Group A and 89% of patients in Group B had positive estrogen receptor (ER) status. There was a trend toward poorly differentiated tumors (Grade 3) in Group A (P = 0.078).
Table 4. Tumor Characteristics
ER: estrogen receptor.
T classification based solely on tumor size. (The parameter ‘skin involvement’ and the T4 category were disregarded, and all tumors were placed in the T1, T2, or T3 category.)
The distribution of disease stages according to current UICC/AJCC criteria after disregarding skin involvement (T4) in favor of tumor size (T1–3) is shown in Table 5. In Group A, cases of extensive disease at regional and distant sites (Stages IIIC and IV, respectively; P = 0.009) predominated. A similarly high level of significance was observed in Group B for the predominance of more favorable Stage I–II disease (P < 0.001). Nineteen patients in Group A (38.0%) and 2 in Group B (7.7%) had distant metastases at presentation. Over the entire study period, distant metastases were diagnosed in 40 patients in Group A (80%), compared with only 9 in Group B (33%). The relative proportions of favorable and unfavorable disease stages clearly demonstrate a trend toward advanced disease in Group A (Fig. 1).
Table 5. International Union Against Cancer/American Joint Committee on Cancer Stage Distribution
Group A (%)
Group B (%)
T classification was based solely on tumor size. (The parameter ‘skin involvement’ and the T4 category were disregarded, and all tumors were placed in the T1, T2, or T3 category.)
In Group A, after a median actuarial follow-up period of 2.5 years (range, 0.1–16.8 years), 44 patients (88%) had died, 36 (72%) due to breast carcinoma and 8 (16%) due to intercurrent disease. At the time of the current analysis, 5 patients (10%) were alive and had no evidence of disease, with a median follow-up duration of 8.7 years. One patient was alive with distant recurrence.
In Group B, after a median follow-up period of 5.3 years (range, 1.4–16.0 years), 15 patients (58%) had died, 7 (27%) due to breast carcinoma and 8 (31%) due to other causes. At the time of the current analysis, 10 patients (38%) were alive and free of disease, with a median follow-up duration of 5.6 years. One patient was alive with distant metastasis.
The clinical outcome of patients with only histologically proven skin involvement (Group B) was significantly superior (P = 0.0003) to that of patients with the classic clinical signs of skin involvement (Group A) (Fig. 2). The 3-year adjusted survival rates were 46.7% in Group A and 92.3% in Group B; the 5-year rates were 38.1% and 83.7%, respectively.
Of the 55 patients without distant metastases at presentation, 18 patients in Group A (58%) and 7 patients in Group B (29%) had disease recurrences at distant sites. Group B was found to have significantly better (P = 0.036) DRFS (Fig. 3). The DRFS rates in Group A and Group B were 63.6% and 91.7%, respectively, at 3 years and 56.9% and 82.0%, respectively, at 5 years.
The mean tumor diameter in Group A was significantly greater than in Group B. Breast carcinoma prognosis is known to become less favorable with increasing tumor size.7 To determine the significance of skin involvement independent of tumor size, a subset analysis of tumors measuring ≤ 3 cm in diameter was performed (Table 6). None of the patients included in this analysis (13 from Group A and 18 from Group B) had distant metastases at initial diagnosis. After a follow-up period of 5 years, decidedly more patients with clinical skin changes died of breast carcinoma (31% vs. 5%). In contrast, more patients from Group B were alive and free of disease or had died of other causes (84% vs. 54%).
Table 6. Subset Analysis: 5-Year Outcome of Patients with Tumor Size ≤ 3 cm
Group A (%)
Group B (%)
Died of metastatic disease
Died of other causes
Alive with metastatic disease
Alive and free of disease
An additional subset analysis was performed to evaluate the seven patients in Group B who died of breast carcinoma (Table 7). The goal was to elucidate whether these patients, independent of a special designation of skin involvement, possessed sufficient predictive factors to indicate the presence of high-risk disease.
Table 7. Subset Analysis of Patients in Group B Who Died of Breast Carcinoma
Age at diagnosis (yrs)
Tumor size (mm)
Lymph node metastases
ER: estrogen receptor; +: positive; −: negative; n.k.: not known.
Defined risk factor.
Patient suffered from severe dementia. Only tumor excision (without axillary lymph node dissection) was performed. Neither irradiation nor adjuvant systemic therapy was considered.
At least 1 of the following prognostic factors had to be present to indicate high-risk status: 1) distant metastasis (Stage IV disease); 2) ≥ 4 positive axillary lymph nodes (N2–3 status); 3) tumor size > 5 cm (T3 status); or 4) the presence of at least 2 of the following 3 additional factors: high histologic grade (Grade 3), negative ER status, and overexpression of HER-2/neu. Using these criteria, five patients were judged to have high-risk breast carcinoma.
The TNM classification system lists macroscopic clinical findings as a criterion for skin involvement in breast carcinoma.4, 5 There was some confusion regarding how to adequately classify cases in which skin involvement is only histologically apparent and does not fulfill the classic clinical criteria (e.g., edema, peau d'orange, ulceration) for the T4b category.6 Due to the absence of strict guidelines, there was a certain amount of leeway in the interpretation of these constellations. At the Institute of Pathology at the University of Basel (Basel, Switzerland), all cases with clinical and/or histologic skin involvement were grouped together in the T4 category until the year 2000. In the second edition of the TNM Supplement, published in 2001, it was decided that only clinically detected skin changes are sufficient for placement in the T4 category.6
There is extensive literature on inflammatory T4 breast carcinoma. The few systematic comparisons between incidence and prognosis for other types of T4 breast carcinoma8–11 exclusively involve patients with apparent clinical disease, and not those with only histologic skin involvement. To our knowledge, the current study is the first to compare patients with histologically proven skin involvement and no corresponding visible changes in the overlying skin (Group B) and patients with classic clinical characteristics (Group A).
Patients in Group A were significantly more likely to have advanced-stage disease. A considerable percentage of patients (38%) already had distant metastases at initial diagnosis. These findings are consistent with those of El-Tamer et al.,11 who reported a 44% rate of distant metastasis at presentation among 99 patients with T4b disease. In contrast, we observed a significantly lower rate (8%) of Stage IV breast carcinoma in Group B.
DSS and DRFS rates were significantly worse in Group A than in Group B. Breast carcinoma with classic clinical skin changes exhibited more aggressive characteristics and different biologic behavior in terms of metastatic capacity. Not only was the proportion of deaths greater in Group A, but death also occurred sooner in this group. The 5-year survival rate of 38% in Group A was similar to data from comparable studies.12–15
It is noteworthy that a cutoff of tumor diameter ≤ 3 cm did not lead to considerable divergence in survival rates between the two patient groups. This finding demonstrates that the clinical morphologic parameter ‘skin involvement’ is an important prognostic factor. Histologically proven skin involvement without corresponding clinical signs does not appear to be a determining factor in outcome. Prognosis, however, is determined primarily by tumor size and lymph node status.16
We did not make further distinctions among the various clinical signs in Group A. Sutherland and Mather10 analyzed 308 patients with regional breast carcinoma (with skin, muscle, and/or chest wall attachment) for the effects of various prognostic factors. Lymph node status and peau d'orange were found to be significant prognostic factors, whereas ulceration and tumor size, among other factors, had no effect on survival.
We therefore conclude that the grouping of all cases with favorable and poor prognosis into one category or stage not only is a theoretic problem (e.g., in that it inevitably leads to incorrect statistical results) but also can pose problems in clinical practice by yielding incorrect information and potentially leading to inappropriate treatment strategies for patients. One of the primary objectives of the TNM system is to simplify communication by creating a universally accepted, coded common language in which the extent of a malignant tumor can be described in ‘shorthand notation’.4, 5 Inherent in the code's simplicity is the danger that a tumor with favorable prognosis will unduly be stigmatized and automatically be placed in an inappropriate category or stage that overstates its malignant potential. Simply due to misleading nomenclature, patients run the risk of receiving overtreatment. In these situations, the intrinsic strength of the system becomes its weakness.
To illustrate this weakness, in the current study, several patients with histologically proven skin involvement and no classic clinical skin changes had tumors measuring < 2.5 cm; these patients underwent mastectomy. In spite of clearly tumor-free margins measuring several centimeters in all directions, these patients underwent irradiation of the chest wall on the grounds that this was the indicated treatment for T4 tumors. The radiotherapy guidelines of the American Society of Clinical Oncology do recommend postmastectomy radiation to combat locoregional recurrence for patients with LABC;17 however, these recommendations are not valid for patients with histologically proven skin involvement, no classic clinical skin changes, and tumor size < 2.5 cm, because the recommendations are based, without exception, on studies of patients with distinctive clinical findings. Currently, no evidence-based data exist regarding postmastectomy radiation for patients with only histologically proven skin involvement, but routine radiation for all patients who incorrectly are considered to have classic T4 disease appears to represent overtreatment. To date, there also are no conclusive data on how to surgically treat patients with only histologic proof of skin involvement. Given the criteria for determining the feasibility of breast-conserving therapy for patients with Stage I–II disease18 and even for those with adequately downstaged LABC,19–21 it should also be possible to use breast-conserving therapy to treat the majority of patients with only histologically proven skin involvement.
We suggest that cases of breast carcinoma with clinical and/or histologic skin involvement can be divided into three groups (Table 8). Cases with clinically and histologically verified skin involvement belong to the T4b/c category;6 this group consists primarily of cases with skin ulceration, because this clinical sign is clearly discernible both macroscopically and microscopically. Cases with skin involvement that can only be verified histologically, with no clinical signs or only discreet ones (e.g., dimpling or retraction of the skin or nipple), do not warrant classification in the T4b/c category; consequently, these tumors are classified as T1, T2, or T3 lesions, depending on their size.6
Table 8. Constellations of Skin Involvement and Their Corresponding T Classifications
+: yes; −: no.
Classification should be based on consensus reached by surgeon and pathologist and should depend on the degree of clinical involvement.
The third group, and the one for which classification is most difficult, includes cases in which clinically visible skin changes are present with no correlating histologic features. This is especially true for clinical features such as edema, peau d'orange, and localized inflammation. These features often are most prominent in the lower half of the breast and periareolar region.22 Rigid characterization of these phenomena is difficult, because characterization depends on the subjective perception of the observer, and also because the characteristics of different phenomena may overlap with each other. The presence of these clinically visible skin changes usually corresponds to involvement of dermal lymphatic channels or obstruction of these channels by the tumor;22, 23 in contrast, there is no pattern of histologic findings specifically associated with clinical diagnosis.24 Histopathologic evidence of dermal lymphatic invasion may be elusive or unconfirmed at the time of pathologic examination,23–25 so it is not clear whether there is a causative relation between dermal lymphatic involvement and clinical features;25 for cases in which histologic evidence is lacking, the current edition of the TNM Supplement recommends that the surgeon inform the pathologist of the clinical findings to ensure that they are considered and, thus, to avoid pathologic understaging.6 Compared with the other recommendations and guidelines of the TNM system, this instruction is uncharacteristically vague. For example, the comparable question of how to manage inflammatory carcinoma is dealt with in a clear and unequivocal manner: if clinical characteristics of inflammatory carcinoma are present (cT4d) but a skin biopsy is negative for tumor and there is a measurable breast malignancy, then the pT category is determined by tumor size.6
It should be noted that each case of clinical skin involvement without histologic confirmation must be evaluated individually. Based on the degree of skin involvement, it must be determined whether a given case can legitimately be categorized as T4 disease. According to our interpretation of the TNM nomenclature, the clinical picture should play the pivotal role in making such a determination.
It is a fundamental principle that although clinical staging may be performed in some cases, pathologic staging is more accurate.26 This tenet, which holds true for smaller tumors (T1–2), loses its validity for LABC. The notion that LABC is primarily a clinical diagnosis, based on findings that usually are readily discernible, also is reflected in the treatment strategies used in the current study population. The current standard of care for patients with LABC, after confirmation of disease by fine-needle aspiration or needle core biopsy, is represented by a multidisciplinary approach, including neoadjuvant or primary systemic therapy, surgery, adjuvant systemic therapy, and radiotherapy.2, 3 Neoadjuvant therapy demands comprehensive clinical staging.
Our results, although possibly influenced by the high median patient age and by the nonuniformity of treatment in the study population, demonstrate that cases of breast carcinoma with only histologic skin involvement and no corresponding clinical features, compared with cases exhibiting advanced clinical spread, are distinct clinical entities and have a significantly different outcome. These features should be emphasized in the nomenclature, so that the recommendation in the TNM Supplement not to classify these cases as T4 disease6 is incorporated into the TNM classification as an actual guideline.4, 5
It has not yet been determined whether cases with histologically diagnosed skin involvement and no corresponding clinical signs should be specially distinguished (e.g., with expanded designations or by subdivision of main categories) in the TNM classification. These cases are quite rare (1–2%), and in our opinion, the clinical significance of skin involvement alone is so minor that telescopic subdivision of the categories T1, T2, and T3 to account for skin involvement would detract unnecessarily from the clarity of the system. Of the few patients in Group B who died of disease, the majority had other predictive factors (e.g., age, lymph node status, tumor size, histologic grade, hormone receptor status, and HER-2/neu status) that indicated their high-risk status.