Presented in part as a platform session at the annual meeting of the United States and Canadian Academy of Pathology, Orlando, Florida, March 1997.
The natural history of patients with intraductal carcinoma (DCIS) and microinvasion is poorly defined, and the clinical management of these patients, with particular reference to management of the axilla, has been controversial. Previous studies of this lesion have used varied and/or arbitrary criteria for the evaluation of microinvasion.
Thirty-eight DCIS lesions with microinvasion (n = 29) or probable microinvasion (n = 9), diagnosed during the period 1980-1996, were retrospectively analyzed after cases not treated with mastectomy and axillary lymph node dissection were excluded. Microinvasion was defined as a single focus of invasive carcinoma ≤2 mm or up to 3 foci of invasion, each ≤1 mm in greatest dimension.
The patients were all females with a mean age of 56.4 years. DCIS was of comedo (n = 31) or papillary (n = 7) subtype. Microinvasion was often associated with an altered, desmoplastic stroma (55%) or a lymphocytic infiltrate (39%). The foci of microinvasion ranged from 0.25 to 1.75 mm (mean, 0.6 mm), with an aggregate mean size of 1.1 mm (range, 0.25-2.25 mm). Foci of microinvasion, ranging from 1 to 3 (mean, 1.7), were adjacent to DCIS in 95.3% of cases. The extent of DCIS did not correlate with the number of foci of microinvasion. Axillary lymph node dissections yielded a mean of 19.3 lymph nodes (range, 7-38), and all lymph nodes were negative for metastasis. None of 33 patients, followed for a mean of 7.5 years (range, 1.0-14.4 years), developed local recurrence or metastasis.
The term "microinvasion," as applied to breast carcinoma, has been used to refer to a highly heterogeneous, variably defined group of neoplasms. Microinvasive ductal carcinoma appears to be a rare lesion, accounting for considerably less than 1% of cases of breast carcinomas seen in our consultation files at the Armed Forces Institute of Pathology; however, in some studies, it has been reported in association with up to 13.5% of cases of intraductal carcinoma (DCIS).1
Given the apparent rarity of microinvasive ductal carcinoma and the lack of an established definition for this pathologic entity, it is not surprising that there is little, and often discordant, data in the literature concerning the biologic behavior of this neoplasm. As such, the surgical management of microinvasive carcinoma, particularly the management of the axilla, has been controversial.1-7
Over the past two decades, the widespread use of mammographic screening for breast carcinoma has resulted in a nearly fivefold increase in the incidence of DCIS8-10 as well as an increase in the detection of early stage invasive carcinoma.11 The biologic significance of these early lesions is not fully understood.11 In this context, using a narrow definition of microinvasion, the current retrospective study was undertaken to characterize the histopathologic features of this entity and the clinical course of patients with this neoplasm.
PATIENTS AND METHODS
The consultation files of the Armed Forces Institute of Pathology were searched for all cases of microinvasive ductal carcinoma of the breast diagnosed between 1980 and 1996. From a total of 15,089 cases of breast carcinoma in our files spanning this 17-year period, 354 cases coded as microinvasive carcinoma or DCIS with early stromal invasion were retrieved. Sixty-seven of these 354 cases were treated with mastectomy and axillary lymph node dissection, and these were selected for review. The relatively large number of "microinvasive" carcinomas in our files is in part a reflection of the consultative nature of our material, with a disproportionately high number of cases posing diagnostic problems. Furthermore, until recently, the term microinvasion was used rather loosely and variably in our department to refer to small invasive carcinomas not exceeding 6-7 mm in greatest dimension.
We have defined microinvasive ductal carcinoma generally as predominantly DCIS with focal invasion of tumor cells into the adjacent stroma. The invasive focus may appear as a tonguelike projection of cohesive neoplastic cells through a disruption in the basement membrane, or may be dissociated from the intraductal component and appear as an infiltration of tumor cells in clusters or singly in the periductal stroma. Although some authors12 have required that such invasive foci extend beyond the terminal duct lobular unit (TDLU) stroma into the interlobular tissue, this is not supported by the ultrastructural confirmation of intralobular capillaries in close proximity to the delimiting fibroblastic layer of terminal ducts.13, 14 Furthermore, DCIS often unfolds the TDLU externalizing the once intralobular stroma. Any distinctive feature of the TDLU stroma is often masked by an inflammatory infiltrate and/or stromal fibrosis. In effect, in the presence of DCIS, the TDLU stroma may not be distinguishable from the interlobular stroma. Quantitatively, the lesion may consist of a single focus of invasion not exceeding 2 mm in greatest dimension, or up to 3 foci of invasion may be present, each not exceeding 1 mm in greatest dimension. Forty-four cases fulfilling these morphologic criteria for microinvasive carcinoma were identified among the 67 carcinomas reviewed. Six cases were excluded because of lack of availability of a complete set of histologic slides. The 38 cases that comprised the final study population included cases of definitive (n = 29) and probable (n = 9) microinvasion (see the next paragraph).
All original hematoxylin and eosin-stained slides from the biopsy and mastectomy (range, 8-28; mean, 16.8) and axillary lymph node dissection (range, 7-38 lymph nodes per dissection; mean, 19.3) were reviewed. Foci of definitive and probable microinvasion were reviewed by two observers. In selected cases, additional levels were examined to assess for the possibility of invasion. Immunohistochemical staining for smooth-muscle actin (mouse monoclonal antibody, 1:4000 dilution, Sigma, St. Louis, MO), the purpose of which was to evaluate for the presence or absence of a myoepithelial cell layer, was performed on paraffin-embedded tissue in 3 cases using the avidin-biotin-peroxidase complex method.15 However, in the majority of cases, the assessment of microinvasion was made solely on the basis of the hematoxylin and eosin-stained slides, as these small lesions were often no longer present when the tissue block was recut in an attempt to evaluate the lesion with additional levels or immunohistochemistry. This accounted for our categorizing some of the lesions as probable rather than definitive microinvasion. The diagnosis of "probable" microinvasion was often made in the setting of an intense chronic inflammatory reaction, in which the possibility of tangential sectioning of DCIS could not unequivocally be distinguished from bona fide microinvasion; however, the latter interpretation was favored. The maximum cross-sectional diameter of each microinvasive focus (to include the intervening stroma separating the invasive cells) was measured on the glass slide to the nearest 0.25 mm using a stage micrometer.
The histologic subtype of the DCIS was recorded. Tumors with a mixed histologic pattern were classified according to the predominant component. Comedo carcinoma was defined by the presence of a highly atypical epithelial proliferation in conjunction with luminal necrosis. In the majority of cases, lesions of comedo DCIS manifested a solid arrangement of tumor cells within ducts; occasionally, cribriform and/or micropapillary patterns were present. The nuclear grade of the intraductal component was graded as 1, 2, or 3 based on the highest grade observed.14 To quantify the amount of DCIS present, the number of cross-sectional and longitudinal duct spaces involved by carcinoma were counted on both biopsy and mastectomy specimens, in a manner similar to that described by Patchefsky et al.16 TDLUs expanded by DCIS (so-called cancerization of the lobule) were counted as single ductal spaces. The extent of DCIS present in a given case was reported as the sum of the number of involved ductal spaces on biopsy and mastectomy slides. To evaluate the relation between the extent of DCIS and the presence of microinvasion, 19 cases of intraductal comedo carcinoma treated with biopsy and mastectomy (n = 17) or mastectomy only (n = 2) were evaluated, as described, to determine the amount of DCIS present. These 19 cases were diagnosed between 1980 and 1994; the mean age of the patients was 57.6 years (range, 45-70 years).
Clinical information was obtained from patients' charts, attending physicians, and tumor registry records. Follow-up information was available for 33 of the 38 patients. The follow-up period was calculated from the date of initial surgery until the most recent follow-up or death. Statistical analysis was performed using Student's t test and analysis of variance.
The 38 patients were all women ranging in age from 32 to 78 years (mean, 56.4 years). Five (13%) were premenopausal at the time of diagnosis. Seventy-six percent of the individuals were white, 11% were African American, and 5% were Asian American; the race was unknown for 8%. The clinical modes of presentation are provided in Table 1. Twenty patients (53%) presented with a palpable mass or nipple abnormality and 18 (47%) with an abnormality on mammogram. Of the latter, microcalcifications with or without an associated abnormal density was most common (56%). Eighty-two percent of the lesions were diagnosed between 1986 and 1996. However, the vast majority (89.5%) of palpable tumors (mean size, 2.2 cm) presented between 1980 and 1990, whereas 88.9% of nonpalpable lesions were detected between 1986 and 1996, corresponding to the increased use of screening mammography. Two patients had undergone modified radical mastectomy for intraductal carcinoma of the contralateral breast diagnosed 1 and 4 years earlier. Another patient had undergone a simple mastectomy for lobular neoplasia of the contralateral breast 2 years previously. One patient had a synchronous contralateral biopsy showing mucinous (colloid) carcinoma; another had a synchronous contralateral biopsy with atypical ductal hyperplasia.
Table 1. Clinical Presentation
No. of cases
Microcalcifications and nonpalpable mass
The patients were treated with modified radical (n = 37) or radical (n = 1) mastectomy. Thirty-six patients (95%) had a diagnostic incisional (n = 1) or excisional (n = 35) biopsy prior to mastectomy. Forty-five percent had 20 or more axillary lymph nodes dissected, and 92% had at least 10 lymph nodes excised. In all cases, axillary lymph nodes were negative for metastatic carcinoma. Chronic lymphocytic leukemia was diagnosed in axillary lymph nodes in one patient; another was found to have non-Hodgkin's (follicular, predominantly small cleaved cell) lymphoma involving one axillary lymph node.
Thirty-three patients were followed for a mean of 7.5 years (range, 1.0-14.4 years). One patient was diagnosed with invasive ductal carcinoma in the contralateral breast 12 months after the diagnosis of microinvasive carcinoma. One patient died of other causes without clinical evidence of recurrent disease at 3.8 years. All other patients were alive without evidence of locally recurrent or metastatic disease at last follow-up.
DCIS was present in all 36 biopsies, and residual DCIS after biopsy was found in 26 (68%) of the mastectomies. Overall, DCIS was present in a mean of 6.5 slides (range, 3-14 slides). The DCIS associated with microinvasion was classified according to the predominant morphologic subtype as either a comedo, Grade 3 lesion or a central papillary, Grade 1 lesion. Comedo DCIS was the dominant morphology in the majority (82%) of cases. It was not possible to address adequately the issue of multicentricity of DCIS (or microinvasion) in this study, largely due to the consultative nature of our material and the variations in information provided to us regarding the specifics of specimen sampling. However, multicentric DCIS was noted in three of eight cases in which sufficient information was available for evaluation. Two of these three patients with multicentric DCIS had concurrent Paget's disease; in all three cases, this was associated with solitary foci of microinvasion.
The typical case was characterized by extensive DCIS of the comedo type with involvement of the majority of TDLUs in a given field. Areas of stromal inflammation or desmoplasia, in which microinvasive foci were present, could often be discerned on low-power examination. The foci of microinvasion consisted of irregular clusters or small aggregates of neoplastic cells, or occasionally isolated tumor cells, devoid of a surrounding myoepithelial cell layer (Fig. 1 (57K)). In the vast majority (95%) of cases, these foci were at a site immediately adjacent to a duct involved with DCIS and often appeared to be dissociated from the intraductal component. In a few cases, the microinvasive foci appeared as a tonguelike projection of tumor through the basement membrane of a duct with DCIS, maintaining continuity with the in situ carcinoma. Microinvasive foci were noted most commonly in association with an altered, desmoplastic stromal reaction (55%), but also in a chronic inflammatory setting (39%) (Fig. 2 (66K)). Microinvasive carcinoma in cases of papillary DCIS more often occurred with a background desmoplastic stroma (77%) than did lesions associated with comedo DCIS (49%). In 4 cases (6%), the microinvasive carcinomas were associated with little or no stromal alteration; 3 of these were lesions that appeared remote from the intraductal component. In each of the latter cases, comedo DCIS was identified elsewhere in the biopsy.
Foci of microinvasive carcinoma were identified in 29 (80%) of biopsy specimens and in 12 (32%) of mastectomies. In 9 of the 12 latter cases, the biopsy contained only DCIS. The majority (87%) of our cases were associated with 1 or 2 foci of microinvasion (Table 2). The mean number of microinvasive foci present per case did not significantly differ among lesions associated with comedo or papillary DCIS (1.6 vs 1.9, respectively; P = not significant [NS]). Each microinvasive focus measured from <0.25 to 1.75 mm in greatest dimension (mean, 0.6 mm), with a mean aggregate size per case of 1.1 mm (range, 0.25-2.25 mm).
Table 2. Extent of Intraductal Carcinoma According to the Number of Foci of Microinvasion
No of foci of microinvasion
0 (n = 19)
1 (n = 15)
2 (n = 12)
3 (n = 4)
DCIS: intraductal carcinoma.
On biopsy and mastectomy combined.
Extent of comedo DCIS (mean no. of involved ductsa)
0 (n = 0)
1 (n = 2)
2 (n = 4)
3 (n = 1)
Extent of papillary DCIS (mean no. of involved ductsa)
Overall, a total of 64 microinvasive lesions were detected in the 38 cases; 50 (78%) were classified as definitive microinvasion and 14 (22%) as probable microinvasion. Lesions associated with comedo DCIS were more frequently diagnosed as definitive microinvasion compared with those occurring in the setting of papillary DCIS (81% vs. 57%, respectively). In part, this may be attributed to prior needle aspiration in some of the latter cases resulting in a periductal stromal tissue reaction and confounding interpretation. Of the 50 lesions that were definitive for microinvasion, 29 (58%) were associated with a desmoplastic stromal reaction and 17 (34%) with a lymphocytic infiltrate. Among the 14 lesions considered to represent probable microinvasion, an inflammatory background was most frequent, present in 8 (57%) of such foci, and accounted for the difficulty in evaluating many of the lesions.
The number of ductal spaces involved by DCIS, on combined biopsy and mastectomy specimens, ranged from 25 to 621 (mean, 247.8). This combined estimate of the extent of DCIS represented an attempt to correct for variations in the sizes of the biopsy specimens. Cases of comedo DCIS associated with microinvasion involved a significantly greater number of ductal spaces (mean, 279.6) than papillary DCIS associated with microinvasion (mean, 106.6) (P = 0.003) and comedo DCIS without microinvasion (mean, 223.2) (P = 0.05). However, the extent of comedo DCIS did not correlate with the number of foci of microinvasion present (P = NS) (Table 2). For papillary DCIS, a trend for greater extent of DCIS with increasing number of microinvasive foci was present, although not statistically significant.
The definition of microinvasive ductal carcinoma is not well established in the literature; consequently, a wide range of diagnostic criteria have been used in published studies for the evaluation of microinvasion. These have encompassed such general descriptive terms as foci of "stromal invasion,"17 "microscopic focus of malignant cells invading beyond the basement membrane,"3 and "one or two minute foci of invasion"18 in addition to quantitative criteria ranging from ≤1 mm1, 19, 20 to ≤5 mm of invasion.10, 21 Other investigators have recognized microinvasion in cases in which invasion comprised <10% of the tumor5 or up to "10% of the surface of the histologic sections."16, 22 Still others have categorized microinvasive carcinoma with a group of diverse lesions commonly referred to as "minimal breast cancer," which includes in situ carcinomas23 as well as carcinomas with invasion up to 5 mm21, 24 or 1 cm.25, 26 Moreover, in several published reports addressing management issues related to microinvasive carcinoma, specific criteria for the diagnosis of microinvasion were not provided.9, 27-29
Interpretation of data from the literature relating to axillary lymph node status and clinical outcome is likely to be inconclusive if meticulous attention is not paid to the diagnostic criteria and methodologies used in the evaluation of microinvasion. It is therefore not surprising that there has been considerable controversy regarding the management of DCIS accompanied by a focus (or foci) of microinvasion.1-7
In the current study, using a stringent and reproducible definition of microinvasive carcinoma, we have characterized the clinicopathologic features and clinical outcome associated with this lesion. Case inclusion in this study mandated treatment with modified radical or radical mastectomy to allow assessment of axillary lymph nodes and provide assurance that occult invasive foci were not present in unresected breast tissue. Because of the limitations imposed by this study design and the consultative nature of our material, a determination of the incidence of microinvasion among our cases was not feasible.
Our definition of microinvasion allowed for a solitary invasive focus of up to 2 mm in greatest dimension or up to 3 foci of invasion, each not exceeding 1 mm in greatest dimension. Utilizing these criteria, our 38 cases of microinvasion had a mean aggregate size of 1.1 mm. Theoretically, even a single neoplastic cell beyond the basement membrane of the reference duct would qualify as microinvasion, although in this study the smallest lesion consisted of a cluster of five to six tumor cells. Although this definition was admittedly arbitrary, it was supported by the demonstration that neoplastic cells in vivo proliferate as avascular spheroids up to only 2 mm,30, 31 constrained by the physical determinants of passive diffusion. Beyond this size, neoplastic growth is angiogenesis dependent.32 Furthermore, 2 mm is the maximum size acceptable as a micrometastasis in axillary lymph node deposits.33 In any case, given the excellent outcome and absence of axillary lymph node metastases associated with this lesion so defined, our definition suggests a minimum requisite size for which use of the term microinvasion is justifiable. The maximum allowable size for a diagnosis of microinvasion was not, however, addressed in this study and warrants further investigation.
Precise measurement of DCIS is often not possible, as many cases present with nonpalpable microcalcifications on mammography and are not associated with a discrete mass. Although several methods have been suggested for assessing the extent or size of DCIS,34 we used the number of involved ductal spaces as an approximation of the extent of DCIS, as it has been shown to correlate best with the mammographic estimate of size. The latter closely parallels the microscopically confirmed measurement of DCIS, particularly for high grade (comedo) DCIS.35, 36 Although our cases of comedo DCIS with microinvasion involved a significantly greater number of ductal spaces than did a comparable group of DCIS without microinvasion, the minimum extent of ductal involvement that excludes microinvasion is not known. In this study, microinvasion was detected in even relatively small lesions of DCIS with only 25 involved ductal spaces. Similar findings were reported by Silverstein et al.18 Using a definition of microinvasion comparable to ours, they showed that the three-dimensional reconstructed size of DCIS correlated with microinvasion, although microinvasion was also present in association with a small lesion of DCIS not exceeding 5 mm. Patchefsky et al.16 described a relation between the incidence of microinvasion and high (>50) duct counts for comedo DCIS, and, of particular interest, found microinvasion in 42% of comedo DCIS involving <50 ductal spaces. We employed a similar methodology for evaluating the extent of DCIS, and none of our microinvasive carcinomas associated with comedo DCIS had <100 involved ductal spaces.
Comedo carcinoma was the most frequent histologic subtype of DCIS associated with microinvasion, accounting for 81.6% of our cases. Comparable findings have been reported by others, with comedo DCIS comprising 66.7-80% of cases of microinvasion.1, 2, 5, 16, 29, 37 This distribution, however, may be skewed by the case selection with a predilection for cases treated with mastectomy and/or axillary lymph node dissection, which may bias toward cases of comedo carcinoma. Papillary DCIS has accounted for 6.2-8.3% of reported microinvasive carcinomas,1, 5, 16, 37 compared with 18.4% in our series. We have seen low grade cribriform DCIS confined to the TDLU with microinvasion (Fig. 3 (50K)); however, such cases were not included in this study because mastectomy was not performed. Likewise, microinvasion may also be seen in association with lobular neoplasia, but this study was restricted to DCIS cases.
Several studies of microinvasive carcinoma have evaluated the clinical presentation of these lesions. Among cases diagnosed between 1965 and 1988, 23-56% presented as abnormalities on screening mammogram (most frequently microcalcifications), and 33-63% were clinically detected as palpable masses.3, 17, 38 This distribution corresponds to that found in the current study, despite the later time frame in which our cases were diagnosed. Several cases of microinvasive carcinoma that initially presented as Paget's disease have been reported;2, 3, 38 these findings were similar to ours.
Although none of our cases were associated with lymph node metastases, the reported incidence of histologically confirmed metastases to axillary lymph nodes in cases of microinvasive (variably defined) carcinoma varies considerably, ranging from 0% to 20% (Table 3). Although our results and those of others3, 6, 19, 40 suggest that microinvasive carcinoma is an extremely favorable lesion, several deaths among lymph node positive patients attributed to carcinoma have been reported.9, 17 (Table 3). However, with one exception, all studies that have reported axillary lymph node positivity in this setting have either allowed a larger maximum size to qualify as microinvasion than that used in the current study or have not even defined microinvasion (Table 3). Silverstein et al.,37 using a definition roughly comparable to ours, reported axillary metastases in one (4.2%) of 24 cases of microinvasion. Unfortunately, because the majority of cited studies1, 3, 5, 9, 17, 37 have not directly addressed the adequacy of sampling of their reported cases, the possibility of occult frank invasion accounting for lymph node metastases cannot entirely be excluded. Given that routine axillary dissection for patients with early stage (T1) breast carcinoma and clinically negative lymph nodes has been questioned in light of the low incidence of lymph node metastases,7, 41, 42 the lack of conclusive therapeutic value,43, 44 the potential for significant morbidity,45 and the exorbitantly high cost-to-benefit ratio,46, 47 additional studies that pay meticulous attention to sampling are indicated to evaluate further the efficacy of routine axillary staging for this subset of T1a tumors.
Table 3. The Status of Axillary Lymph Nodes in Published Cases of Microinvasive Carcinoma
Definition of microinvasion
No. of cases
No. of cases with positive axillary lymph nodes (%)
No. of positive lymph nodes/dissection
Follow-up (lymph node positive cases)
DOD: dead of disease; AWR: alive with local recurrence; NED: no evidence of disease; NA: not available.
The patient had synchronous 1.0 cm invasive carcinoma outside the biopsy quadrant.
It is not known whether all 25 patients underwent axillary lymph node dissection.
Axillary lymph node dissection was performed on 33 of 41 patients.
1 focus of invasion ≤2 mm in greatest dimension, or up to 3 foci, each ≤1 mm in greatest dimension
The optimal management of the breast in patients with DCIS and microinvasion was not specifically addressed in this study. There is, however, some concern regarding breast-conserving surgery in this setting because of the risk of occult invasive carcinoma, as reported by Rosner et al.4 Defining microinvasion as "limited microscopic stromal invasion ... not invading more than 10% of the surface of the histologic sections," they detected frankly invasive residual carcinoma in the mastectomy specimens of 6 (18%) of 33 cases with microinvasion on excisional biopsy.
In our 33 cases, having a mean follow-up of 90 months, there were no recurrences or deaths from carcinoma. Very few studies of DCIS with microinvasion that have used a definition of microinvasion roughly comparable to ours have also provided follow-up data on the patients. Wong et al.3 and Rosner and Lane38 reported 100% disease free survival rates with follow-up periods of 47 months (median) and 99 months (mean), respectively. Silverstein48 described recurrences in 2 (9%) of 21 cases with 100% survival during a median follow-up of 85 months. Based on these small studies, it appears that the natural history of DCIS with microinvasion more closely parallels DCIS than invasive carcinoma. Longer follow-up of a larger series of patients is indicated to assess adequately the prognostic significance of this lesion.
The opinions and assertions contained herein are the private views of the authors and are not to be construed as official or as representing the views of the Department of the Army or the Department of Defense.