Microinvasive breast carcinoma

Clinicopathologic analysis of a single institution experience




Microinvasive breast carcinoma (MIC) has a good prognosis but specific definitions have varied in the past, making the clinical significance of MIC a subject of debate.


Microscopic slides of 59 cases of breast carcinoma originally diagnosed as MIC were reviewed retrospectively. Histologic parameters were correlated with clinical findings and outcome to define diagnostic criteria better.


On review, the 59 cases were recategorized as follows: pure DCIS (N = 16), DCIS with foci equivocal for microinvasion (N = 7), DCIS with ≥ 1 focus of microinvasion (N = 11), T1 invasive carcinomas with ≥ 90% DCIS (N = 18), and T1 tumors with < 90% DCIS (N = 7). The MIC cases in the current study averaged 3 separate foci of early infiltration outside the basement membrane, each one not > 1.0 mm. The mean follow-up was 95 months. Six patients (10%) had only local recurrence: 1 case each in patients with equivocal microinvasion, microinvasion, and T1 tumors with < 90% DCIS and 3 cases among the patients with T1 tumors with ≥ 90% DCIS. Four patients, all with T1 tumors with ≥ 90% DCIS, had distant failure (7%). In the MIC group, only one patient developed a local recurrence after breast conservation. No patient had axillary lymph node metastasis. For the entire series, factors associated with local recurrence were younger age, breast conservation versus mastectomy, and close surgical margins. The only factor associated with distant failure was the size of the DCIS component. Seven patients with T1 tumors with ≥ 90% DCIS experienced local or distant failure and 5 of these (71%) developed progressive disease or died of disease. All other patients who developed a recurrence were disease free at last follow-up. In a retrospective series, poorer outcome in carcinomas with ≥ 90% DCIS may be related to the greater likelihood of missed larger areas of invasive carcinoma. Therefore, meticulous and extensive sampling of these carcinomas is required.


MIC as defined has a good prognosis. It has a different biology than T1 invasive carcinoma with ≥ 90% DCIS, which may progress and cause death. Large tumors with multiple foci of microinvasion may have metastatic potential. Cancer 2000;88:1403–9. © 2000 American Cancer Society.

Carcinoma of the breast increasingly is being diagnosed at an earlier stage because of routine screening mammography, with more cases being diagnosed at the critical juncture between in situ and invasive carcinoma. There have been many attempts to identify such cases as microinvasive carcinoma (MIC), but definitions have varied widely, making the usefulness of distinguishing MIC as a distinct category of breast carcinoma open to question.1 At one end of the spectrum is the definition of MIC given by Silverstein et al.,2 i.e., with 1 or 2 microscopic foci of possible invasion not > 1 mm in greatest dimension. A similar definition has been used more recently by Silver and Tavassoli3 that categorizes MIC as a single focus of invasive carcinoma (IC) < 2 mm or up to 3 foci of invasion, each ≤ 1 mm in greatest dimension. Additional definitions occur in the literature with this general philosophy, not all of which limit the allowable number of invasive foci.4–8 At the other end of the spectrum of definitions is that of Solin et al.,9 which defines MIC as the maximal extent of IC comprising < 10 % of the tumor, with ≥ 90% ductal carcinoma in situ (DCIS). These varying definitions of microinvasion (e.g., such as that of Wong et al.10) have been a source of nonuniformity in the analysis of the clinical outcome, which has led to uncertainty regarding the separation of MIC from DCIS on the one hand and, conversely, from small ICs. In an attempt to test the efficacy of the various definitions of MIC, we reviewed cases originally diagnosed as MIC at our institution and stratified the degree of invasion into definable groups to correlate each histologic group with clinical outcome. Although such stratification resulted in fewer cases per group and hampered statistical analysis, this allowed us to better observe the biology of early invasive breast carcinoma.


The clinical history and pathology material of 67 patients diagnosed with MIC at the Fox Chase Cancer Center (FCCC) between 1982–1992 were reviewed from the records of the tumor registry and the Departments of Radiation Oncology and Surgery. Cases were approximately evenly divided between patients first biopsied at other institutions and those whose first surgery was at FCCC. Informed consent was obtained from all subjects. All pathology materials and all available slides were reviewed by two pathologists (R.F.P. and A.S.P.) together and a consensus diagnosis was reached. As expected, these cases proved to be a mixed population of tumors ranging from DCIS to fully invasive ductal carcinoma (IC). Eight of the 67 cases were excluded from further analysis because of the following reasons: unavailability of slides for pathologic review (5 cases), lobular carcinoma in situ (LCIS) only (1 case), lack of clinical follow-up (1 case), and 1 case that on histologic review was interpreted as a 1.2-cm IC.

An average of 16 microscopic slides of breast tissue were reviewed for each case, (range, 2–41 slides). Fewer slides were available, on average, from 13 cases comprised entirely of outside slide review material (average, 10 slides per case) versus 46 cases with at least 1 surgical procedure performed at FCCC (average, 18 slides per case). Tumor size, particularly in this retrospective series, was difficult to measure. As suggested by Lagios,11 we attempted to arrive at the best possible estimate of tumor size by utilizing all available information, including mammographic size, macroscopic pathologic size, size measured from histologic sections, and the number of consecutive histologic slides involved. The size of the DCIS and IC were measured separately to conform with current TNM guidelines. Nuclear grade of DCIS and IC was estimated using previously published criteria.12, 13

For patients treated with breast-conserving surgery, tumor margins were assessed microscopically from the available slides using an ocular micrometer. Tumor > 2 mm away from the inked margin was considered a negative margin. Tumor painted by ink was considered a positive margin. Tumors in between these two confines were considered close. In some cases, the tumor margins were assessed macroscopically to be negative but no histologic sections were taken, or the slides of a reexcision specimen were not available for review but the surgical pathology report indicated the margins were negative. These were considered as negative by macroscopic examination or by report but were unconfirmed histologically. In some cases, neither the slides nor the report of reexcision specimens were available for review, or no assessment of tumor margins was made; these cases were considered as indeterminate. Five of 13 cases (38%) comprised entirely of outside slide review material had histologically verifiable margins. Tumor margins were assessed histologically in 35 of 46 cases (76%) with at least 1 surgical procedure performed at FCCC.

The range of tumors analyzed in this study appears to represent a morphologic continuum from DCIS to small ICs that straddle the variety of definitions of microinvasion that have been proposed. The groupings selected in this study were chosen to supply nonoverlapping categories that could be compared and related readily to the American Joint Committee on Cancer staging system (TNM),14 as well as taking into account the extent of the in situ and invasive components (Table 1).

Table 1. Number of Patients in Each Histologic Group
HistologyNo. of patients (%)
  1. DCIS: ductal carcinoma in situ.

Pure DCIS16 (27%)
Equivocal for microinvasion7 (12%)
Early microinvasion5 (8%)
Late microinvasion6 (10%)
T1 tumors, ≥ 90% DCIS18 (31%)
T1 tumors, < 90% DCIS7 (12%)
59 (100%)

Tumors were segregated as showing no invasion (pure DCIS). Another group of tumors predominantly was DCIS, but showed focal penetration or breaching of the ductal perimeter by a few cells that appeared disassociated and unattached to the epithelial basement membrane (early microinvasion) (Fig. 1B). These foci always were accompanied by a stromal response comprised of scattered chronic inflammatory cells arranged within pale staining loose arrays of new collagen. In addition to these qualitative aspects, the total number of such areas was counted for each case.

Figure 1.

(A) Ductal carcinoma in situ (DCIS) with foci equivocal for microinvasion (H & E, ×175). (B) DCIS with early microinvasion (H & E, ×350). (C) DCIS with late microinvasion (H & E, ×350). (D) Small invasive ductal carcinoma (T1a). The breast lobule shows both DCIS (left) and invasive carcinoma (right) (H & E, ×70).

There also were cases that showed the characteristic periductal stromal reaction of early microinvasion, but the presence of detached epithelial cells outside the basement membrane was a matter of debate or equivocal (Fig. 1A). These cases were segregated into a separate category as showing “equivocal microinvasion” for the purpose of further analysis.

Late microinvasion (Fig. 1C) was defined as a progression of early microinvasion. These predominantly were DCIS, but showed areas of definite stromal infiltration by groups of more than a few loosely cohesive carcinoma cells that remained close to the epithelial basement membrane but infiltrated to a greater degree than that observed in early microinvasion. None of these invasive areas measured > 1.0 mm in greatest dimension as viewed with the ocular micrometer. These cases also were analyzed quantitatively for the number of such invasive foci, similar to those in early microinvasion.

The next group was comprised of ICs > 1 mm (T1; range, 0.15–0.9 cm) comprised of a maximum of 10% IC and thereby comprised of at least 90% in situ carcinoma as judged from the areas of tumor involvement on the microscopic slides (Fig. 1D). Also included in this group was 1 exceptional case of an 8-cm DCIS with 17 foci of microinvasion.

The final group was comprised of IC > 1 mm (T1; range, 0.3–0.9 cm), in which the invasive component comprised > 10% of the tumor.

Patients treated by either breast conservation (lumpectomy and radiation therapy) or mastectomy were studied. Follow-up of all cases was obtained from patient charts, the files of the tumor registry, and the data bank of the Department of Radiation Oncology. Overall survival, cause specific survival, and recurrence free survival were estimated using Kaplan-Meier methodology. A local regional recurrence was defined as a recurrence in the ipsilateral breast, chest wall, or axillary lymph nodes. All other recurrences were considered as distant. Associations between recurrence, age, and histologic parameters were evaluated using the chi-square test.


Review of the study group of 59 breast carcinomas originally labeled as MIC resulted in the tumors being reclassified as follows: 16 cases of pure DCIS, 7 equivocal for microinvasion, 5 microinvasive with focal breach of the basement membrane (early microinvasion), 6 microinvasive ≤ 1 mm (late microinvasion), 18 invasive carcinomas of T1 (> 1 mm) with ≥ 90% DCIS, and 7 invasive carcinomas of T1 (> 1 mm) with < 90% DCIS (Table 1). Within the T1 invasive tumors with ≥ 90% DCIS, 1 case was T1b IC and 1 case was a large, 8-cm DCIS with 17 foci of microinvasion; the remainder all were T1a ICs. The group of T1 invasive tumors with < 90% DCIS was comprised of 3 cases with T1a ICs and 4 cases with T1b ICs.

Analysis of MICs

The total number of MICs was 11. During the same time period as these cases, 1610 patients with breast carcinoma were treated at FCCC for an estimated incidence rate of MIC of 0.68%. The clinicopathologic features of the 11 cases of MIC are presented in detail in Table 2 and are summarized in Table 3. Surprisingly, in this small group, there were three cases of contralateral breast carcinoma (one case each of DCIS, medullary carcinoma, and IC, not otherwise specified) and three cases of other malignancies (one each of melanoma, colon carcinoma, and malignant mixed mesodermal tumor of the ovary). A history of breast carcinoma in first-degree relatives was elicited in 4 patients (36%). It is interesting to note that two patients with MIC had both contralateral breast tumors as well as other malignancies. The only death among the MIC cases was the patient in Case 1, who died of malignant melanoma with no evidence of recurrent breast carcinoma.

Table 2. Actual Microinvasive Breast Carcinoma Cases (Including Early and Late Microinvasive Carcinomas)
Case no.Age (yrs)PresentationTreatmentSize DCIS (cm)Grade DCISHistologic type DCISNo. foci microinvMarginsTumor in reexOutcome (mos after diagnosis)
  1. DCIS: ductal carcinoma in situ; microinv: microinvasions; reex: reexcision; Mam: mammographic; MRM: modified radical mastectomy; neg: negative; exam: examination; DOC: dead of other causes; Lump: lumpectomy; Inter: intermediate; Pap: papillary; NED: no evidence of disease (months of follow-up); Clin: clinically palpable; excor: excoriation; histol: histology; micropap: micropapillary; recur: recurrence; crib: cribriform; MMMT: malignant mixed mesodermal tumor.

181MamMRM2.5HighComedo3Neg macroscopic examDOC (85)
267MamLump + reex1.4InterPap/solid3Close DCISResidual DCIS + maybe microinvNED (101)
336Clin-excor nippleMRM2.0HighComedo3Neg histolNED (94)
441Clin-massMRM1.5HighComedo6Neg histolNED (102)
569Clin-massMRM3.6HighComedo1Neg histolNED (98)
670MamLump1.3InterSolid1Neg macroscopic examNED (130)
748Clin-massLump + reex3.0HighMicropap2Neg reportResidual, type unknownLocal recur DCIS (24), NED (26)
840Clin-massMRM6.1HighCrib/micropap/apocrine6Neg histolNED (147)
956Clin-massLump2.8HighComedo3Neg histolNED (alive with progressive MMMT) (60)
1047MamLump + reex2.5HighComedo1Neg histolNo residualNED (121)
1149MamLump1.2HighComedo1IndeterminateNED (124)
Mean552.53 foci(99 months)
Table 3. Characteristics of Microinvasive Breast Carcinoma
CharacteristicsMicroinvasive breast carcinoma
  1. XRT: radiation therapy.

Mean age (yrs) (range)55 (36–81)
First-degree relative with breast carcinoma4/11 (36%)
Contralateral breast carcinoma3/11 (27%)
Other carcinoma3/11 (27%)
Presentation: clinical (mean age)6/11 (55%) (48 yrs)
Presentation: mammographic (mean age)5/11 (45%) (63 yrs)
Lumpectomy + XRT (mean age)6/11 (55%) (56 yrs)
Mastectomy (mean age)5/11 (45%) (55 yrs)
Positive lymph nodes0/11 (0%)

Surgical management was nearly equally divided between lumpectomy (six cases) and mastectomy (five cases). All 11 MIC patients underwent axillary lymph node dissection; none had positive lymph nodes. All patients who underwent lumpectomy received radiation therapy, whereas none of the patients undergoing mastectomy did.

The average size of the DCIS was 2.5 cm (range, 1.2–6.1 cm). In the majority of cases the DCIS was high grade (9 of 11 cases; 82%). Comedo-type DCIS was the most common architectural pattern observed (7 of 11 cases; 64%). LCIS accompanied MIC in 1 case (9%). The mean number of foci of microinvasion was three (range, one to six). Approximately 67% of cases had ≥ 2 foci of microinvasion. Three of 6 patients treated by lumpectomy and radiation therapy underwent reexcisions as a separate surgical procedure; of these, 2 patients (67%) had residual tumor present in the reexcision specimens. Final margins were negative histologically or by report in 9 of 11 patients (82%), indeterminate in 1 patient (9%), and histologically close in 1 patient (9%). One of six patients treated with breast conservation developed a local recurrence (Case 7). The reportedly negative surgical margins in this case could not be confirmed by histology because the slides were not available for review.

Analysis of Treatment Failures

Of the 59 breast carcinomas retrospectively reviewed, 6 patients developed local recurrences, 3 patients developed distant metastases, and 1 patient developed both local and distant recurrences.

Only one of the recurrences occurred among the MIC cases. This was a local recurrence of DCIS (Case 7 in Table 2). This patient had histologically unverified tumor margins in the original specimen (slides not available for review), and was treated with salvage mastectomy. At last follow-up this patient was without evidence of disease, 2 months after the recurrence.

There were no recurrences in the 16 patients with pure DCIS. One patient with equivocal microinvasion, 3 with T1 tumors with ≥ 90% DCIS, and 1 with T1 tumor with < 90% DCIS developed local recurrences. Of these five local recurrences, three had close margins and one had margins verified by macroscopic examination only. Only one of the locally recurrent cases had histologically verified negative tumor margins. The only uncontrolled local recurrence occurred in a patient with a T1 tumor with ≥ 90% DCIS. All three patients with distant metastases and the single patient with both a local and distant recurrence presented with T1 tumors with ≥ 90% DCIS. One of these patients died of disease and the remaining two patients showed progression of their metastatic disease at last follow-up.

Only 1 of the entire study group of 59 patients had lymph node metastasis at presentation. This patient had a T1a invasive tumor with ≥ 90% DCIS and 4 of 7 positive axillary lymph nodes. After mastectomy, chemotherapy, and hormonal therapy, the patient was without evidence of disease at last follow-up.

Two deaths from causes unrelated to the initial breast carcinoma occurred within the study group of 59 patients during follow-up. Neither of these patients manifested recurrence of their original breast tumors.

Patients who developed local or distant recurrences were significantly younger at the time of diagnosis than patients without recurrence. The mean age at the time of initial diagnosis was 48 years for all patients with local recurrence and 45 years for all patients with distant failures, which was approximately 10 years younger than the mean age at presentation of the group as a whole (57 years) (P < 0.03).

Both local and distant recurrence was related significantly to the size of the DCIS and not to the size of the invasive component. The mean size of the in situ component in the locally recurrent cases (2.7 cm) and distant failures (5.6 cm) was greater than that of the study group as a whole (2.2 cm) (P < 0.001). The size of the invasive component was not found to be related to recurrence.

In the entire study, only 1 of 30 patients (3%) with histologically verified negative resection margins (> 2 mm) developed a local recurrence (P < 0.01). Although all patients with local failure only initially were treated with breast-conserving therapy, the local failure rate can be ascribed to the tumor margin status rather than the treatment modality. Similarly, all patients who failed distantly underwent mastectomy as their initial treatment.

Actuarial survival for MIC and for the other histologic groups is presented in Table 4. For this study, we chose to include in the T1 tumors with ≥ 90% DCIS group the case of an 8-cm DCIS with 17 foci of microinvasion and pulmonary metastasis. In our opinion, this case clearly falls outside the concept of MIC.

Table 4. Actuarial Survival
Histologic groupNo. of patients5-yr overall5-yr cause specific5-yr recurrence free
  • DCIS: ductal carcinoma in situ.

  • a

    Groups were too small to calculate 5-year recurrence free survival.

Pure DCIS16100%100%100%
Equivocal microinvasion7100%One regional recurrencea
T1 tumors, ≥ 90% DCIS18100%100%81%
T1 tumors, < 90% DCIS7100%One regional recurrencea


MIC is an uncommon histologic subtype of breast carcinoma. The referral service of the Armed Forces Institute of Pathology reports an incidence rate of 2.4%.3 The incidence rate in the current series, based on cases treated at FCCC, an oncologic center, is 0.68%. The actual rate of incidence in the general population most likely is even lower. MIC presents over a wide range of ages; other studies using comparable definitions of MIC show a range of 32–84 years at presentation, with the average age in the sixth decade of life (range, 50–59.8 years).3, 15, 16 This is similar to the current series, in which patients presented at an average age of 55 years (range, 36–81 years). Recent series using similar definitions report that MIC presents as a palpable mass in 19–50% of cases.3, 15, 16 This is similar to our finding that 45% of patients in the current study presented with a palpable mass. MIC has been found to be more likely to present with a palpable mass than DCIS,15 but a comparable comparison group was not part of the current study. MIC most often is associated with comedo-type or other high grade DCIS, and usually has more than one focus of microinvasion.2, 3, 15, 16 In the current series 9 of 11 patients (82%) had high nuclear grade, and the average number of invasive foci was 3. Thus our cases fit the general clinicopathologic profile of other recent series using comparable definitions.3, 15, 16 The occurrence of bilateral breast carcinoma, ICs at other sites, and 33% of patients with a first-degree relative with breast carcinoma may suggest a possible common oncogenic factor, with MIC as an expression of the phenotype. Further investigation of this possibility may be informative. An example of this already occurs in breast carcinoma, in the association of poorly differentiated, estrogen receptor negative carcinoma with the BRCA1 mutation.17, 18

MIC appears to have a low rate of recurrence; some investigators have found no recurrences,3, 10 whereas Silverstein15 found 2 cases of recurrence (2 of 21 cases; 9.5%) after a median follow-up of 85 months. Our own experience shows that 1 recurrence occurred in our group of 11 MICs. This local recurrence developed in a patient with histologically unconfirmed tumor margins. Inadequate local control also accounted for a single local recurrence in Silverstein's series of MIC.15 The importance of obtaining histologically verified negative tumor margins also is highlighted by the one case in the current study with equivocal microinvasion that locally recurred but had close histologic margins.19 This latter group of cases would be included in some studies of MIC, but we have chosen to analyze them separately to discern any differences in the behavior of a spectrum of cases starting with the earliest possible evidence of tumor invasion.

The one patient with 17 foci of microinvasion who developed lung metastases most likely would be excluded from studies of MIC using current criteria20 and in our study we consider this a T1 tumor with ≥ 90% DCIS. In recent study, Zavotsky et al.21 described a similar case of MIC with multiple foci suspicious for microinvasion that were < 1 mm. A single sentinel lymph node was positive for a 0.3-cm metastasis. The patient developed a malignant pleural effusion 29 months after breast surgery. Therefore, evidence is accumulating in the literature that cases of DCIS with multiple foci of microinvasion have significant metastatic potential and should be distinguished from MIC without multiple foci of microinvasion.

Axillary lymph node metastasis in MIC has been reported in a minority of cases (range, 3–18%)9, 16, 22–25 or not at all,3, 10, 15 depending on the definition of MIC used in each study. A recent study of MIC using sentinel lymph node mapping found 14.3% of patients (2 of 14) with lymph node metastases.21 None of the patients with MIC in the current study had lymph node metastases. Silverstein15 also noted no lymph node metastases in 17 cases. In comparison, 2 of 321 lymph node dissections performed for DCIS were positive (0.6%), which emphasizes the need to compare the treatment and outcome of MIC with DCIS and not normal controls.15

Seven failures in the current study, both local and distant, occurred in patients with invasive (T1) carcinoma with ≥ 90% DCIS. Failure within this group appeared to have a poor prognosis because 5 of 7 patients (71%) had progressive disease or died of disease. In IC, studies have found a higher rate of local recurrence26 and metastatic disease27 in cases in which the invasive breast carcinoma is associated with an extensive intraductal component and the patient is young.25, 28 Both of these features were associated with treatment failure in the current series. Such tumors should not be included in any definitions of MIC.

Treatment failure was associated with the size of the DCIS rather than the size of the invasive tumor. Because the current study is a retrospective series, this might be due to the fact that tumors bearing the larger burden of DCIS might not have been sampled enough to detect areas of frank invasion.

MIC, as defined by Silverstein et al.2 and Silver and Tavassoli,3 has a good prognosis. In the current series no cases with an average of 3 separate foci of ≤ 1.0 mm microinvasion recurred, unless tumor margins were unconfirmed histologically. Cases of MIC with multiple foci of microinvasion may identify a group with metastatic potential, although the threshold number of such foci is unknown. For this reason, we would recommend that all tumors that show a focus of invasion ≤ 1 mm be examined thoroughly to document the number of such foci in the tumor, as well as the possibility of larger areas of invasion. At this time, MIC as defined by Silverstein et al.,2 Silver and Tavassoli,3 and the American Joint Committee on Cancer staging system (TMN)14 should be adhered to if the term is to have any clinical relevance. A multiinstitutional study with the accrual of more cases studied uniformly may cast light on the question of the threshold number of microinvasive foci that will predict more aggressive behavior. The current study has shown that MIC should be distinguished from T1 invasive tumors with ≥ 90% DCIS. These latter tumors have aggressive potential and are associated with a poor outcome, especially when occurring in younger women. Such cases illustrate that there is no place for any definition of MIC based only on the percentage of DCIS in the tumor.