Automatic segmentation of breast MR images through a Markov random field statistical model

An algorithm dedicated to automatic segmentation of breast magnetic resonance images is presented in this paper. Our approach is based on a pipeline that includes a denoising step and statistical segmentation. The noise removal preprocessing relies on an anisotropic diffusion scheme, whereas the statistical segmentation is conducted through a Markov random field model. The continuous updating of all parameters governing the diffusion process enables automatic denoising, and the partial volume effect is also addressed during the labeling step. To assess the relevance, the Jaccard similarity coefficient was computed. Experiments were conducted on synthetic data and breast magnetic resonance images extracted from a high-risk population. The relevance of the approach for the dataset is highlighted, and we demonstrate accuracy superior to that of traditional clustering algorithms. The results emphasize the benefits of both denoising guided by input data and the inclusion of spatial dependency through a Markov random field. For example, the Jaccard coefficient for the clinical data was increased by 114%, 109%, and 140% with respect to a K-means algorithm and, respectively, for the adipose, glandular and muscle and skin components. Moreover, the agreement between the manual segmentations provided by an experienced radiologist and the automatic segmentations performed with this algorithm was good, with Jaccard coefficients equal to 0.769, 0.756, and 0.694 for the above-mentioned classes.

Neoadjuvant endocrine therapy in breast cancer

The clinical benefits of endocrine therapy for patients with hormonosensitive breast cancer are well established. For many years, five years' treatment with tamoxifen was the gold standard of adjuvant treatment. The recent development of new endocrine agents provides physicians with the opportunity to take a more effective therapeutic approach. Nevertheless, the success of neoadjuvant endocrine therapy is much more recent and less frequently reported in the literature. This article reviews the studies published on neoadjuvant endocrine treatment (tamoxifen and aromatase inhibitors). According to the literature, neoadjuvant endocrine therapy seems to be effective and well tolerated. The newer generation of aromatase inhibitors (letrozole, anastrozole, exemestane) appear to result in better overall response rates and more conservative surgery than tamoxifen. Patients with an ER Allred score of 6 and over are most likely to respond and gain clinical benefit. The optimal duration of neoadjuvant therapy has not yet been investigated in detail. These preliminary results are interesting and should be confirmed by further studies.

Cancers canalaires in situ du sein avec micro-invasion

OBJECTIVE

Specify the role of axillary lymph node removal in micro-invasive in situ duct carcinomas (DCIS-MI) of the breast with a series of 107 consecutive cases.

METHODS

Single-centre, retrospective, anatomoclinical study with application of the European guidelines adopting as pathological definition the presence of areas of micro-invasion not exceeding 1mm. Axillary lymph node dissection was systematically complete and was preceded by the search for the sentinel node in 10 patients using the isotope method.

RESULTS

Lymph node invasion was revealed in 8 cases in the global population (7.5%). All the cases except one exhibited typical deleterious histological features: comedo architectural sub-type, high nuclear grade, and size of the lesion>3 cm. The lymph node invasion was of 18.5% in the sub-group of high-grade micro-invasive comedo-carcinomas measuring more than 3 cm.

CONCLUSION

Contrary to those exhibiting pure DCIS, DCIS-MI patients require surgical exploration of the armpit, the most appropriate modalities of which are currently debated: classical axillary lymph node dissection or search for the sentinel lymph node; the selective lymphadenectomy procedure is not yet a consensually validated technique.

[Breast cancer detected by a stellar opacity. Histologic characteristics: 155 cases]

Retrospective study (1985-1998) concerning surgical and histopathological features of 155 subclinical breast cancers revealed by spiculated opacity on screening mammograms.

MATERIALS AND METHODS

The patients were 44-78 years old (mean age, 58.5 years), 129 were postmenopausal. Preoperative localization was stereotactic in 57 instances (36.5%), sonographic in 98 instances (63.5%). Maximum tumor diameter varied from four to 25 millimeters (mean diameter 11 mm), below 10 min in 97 cases, below 5 mm in 15 cases. Axillary lymph node dissection was performed immediately (95%) or secondarily (5%).

RESULTS

Subclinical breast tumors exhibiting spiculated picture were infiltrating carcinomas: Infiltrating ductal carcinoma (IDC) in 130 cases (84%), infiltrating lobular carcinoma (ILC) in 25 cases (16%). Not any ductal carcinoma in situ (DCIS) was detected by such an irregular opacity. The grading according to Scarff-Bloom-Richardson was used in IDC: 95 grade I (73%), 31 grade II (24%), four grade III (3%). Hormone receptor status was obtained upon 145 tumors: both estrogen receptors were present in 125 cases (86%). Axillary lymph node involvement (N+) was found in ten cases (6%), always concerning IDC > 5 mm. Conservative surgery was achieved in mort cases (97%).

DISCUSSION

Subclinical breast cancers revealed by spiculated opacity were predominantly corresponding to infiltrating process IDC or ILC, in contrast with breast cancers revealed by microcalcifications, mainly meaning DCIS. In our experience mammographically-detected spiculated malignant tumors were not bearing unfavourable pathological or biological features: they appeared commonly well differentiated and hormonosensitive, furthermore axillary lymph nodes were rarely involved.

[Surgery for subclinical breast lesions]

Subclinical mammographic abnormalities discovered through screening are subjected to excisional surgical biopsy in order to prove or rule out malignant disease of the breast by histological examination. Close attention should be paid by the surgeon to avoid cosmetically disfiguring scar and failure in excision removal exhibited on postoperative mammogram. Indications for surgical assessment are selected upon suspicious patterns of opacities, clusters of microcalcifications or architectural distortions. Good communication between the radiologist, the surgeon and the pathologist is required for proper management of non palpable lesion. The most adequate lumpectomy technique should be performed with regard to the oncologic risk and cosmetic outcome.

Clinical and pathological response to primary chemotherapy in operable breast cancer

Neoadjuvant chemotherapy is used to improve patients' survival in locally-advanced and inflammatory breast cancer and to increase conservative surgical procedures in bulky tumours. Pathological complete responses are unusual. The aim of this pilot study was to assess the clinical and pathological response rates and to evaluate toxicity with a new protocol of primary chemotherapy in 50 high-risk breast cancer patients. All tumours were > 3 cm and had at least one other adverse prognostic factor: lymph node involvement (32 N1, 6 N2), SBR grade III (20), aneuploidy (29), negative hormonal receptors (19). Patients were treated by 3-week cycles of THP-doxorubicin 20 mg/m2 D1 to 3, vinorelbine 25 mg/m2 D1 and 4, cyclophosphamide 300 mg/m2 and 5-fluorouracil 400 mg/m2 D1 to 4 (TNCF). 38 patients received G-CSF or GM-CSF support. After 4-6 cycles, all underwent surgery (39 conservative, 11 modified radical). Tumour response was assessed clinically, by mammography and echography and on pathological specimens. An objective clinical response was observed for 43 patients: 26 complete (51%) and 18 partial (37%). After pathological review, 11 patients (22%) were devoid of any tumour cells, 4 others (8%) had only in situ carcinoma. From 253 evaluated cycles, grade III-IV toxicity occurred, 81% with neutropenia, 25% with anaemia, and 20% with thrombocytopenia. All patients recovered. This regimen induced a severe but not life-threatening haematological toxicity and resulted in a high pathological response rate (30%).

[First-line chemotherapy in operable breast neoplasms: results of 3 protocols]

In order to avoid modified radical mastectomy, a neoadjuvant approach was adopted in our institute for operable bulky breast cancers. From January, 1982, to December, 1995, 288 patients received primary chemotherapy with 3 different regimens (all doses mg/m2): (1) AVCF/AVCFM, 167 patients (adriamycin 30, vincristine 1 d1, cyclophosphamide 300, fluorouracil 400 d2-d5 and methotrexate 20 d2 and d4, every 28 days); (2) NEM, 78 patients (vinorelbine 25, epirubicin 35, methotrexate 20 d1 and d8, every 28 days); and (3) TNCF, 43 patients (THP-adria 20, d1-d3, vinorelbine 25 d1 and d4, cyclophosphamide 300, fluorouracil 400 d1-d4, every 21 days). Evaluation of the response comprised 3 methods: clinical (C), echographic (E), mammographic (M). The overall objective response rate (C: 63/90/93; E: 49/61/85; M: 53/65/83%) is higher with regimens (2) and (3). The complete response rate was increased 2-fold with TNCF but the hematologic toxicity was very superior with this combination. Patients were all operated for (2) and (3), only several for (1), and the breast conservation rate (68/83/79%) was quite similar in the 3 regimens. The pathological complete response rate reached 23% with TNCF. However the impact on patient survival has to be confirmed.

[Axillary lymph node dissection in clinically occult breast cancer]

The study concerns 265 patients with axillary lymph node dissection for non-palpable breast cancer. The mammographically detected breast tumors were: 36 ductal carcinomas in situ (DCIS), 23 microinvasive carcinomas, 206 invasive carcinomas of which 179 were invasive ductal cancers (IDC), 25 invasive lobular cancers (ILC) and 2 mucinous invasive carcinomas. The histologic size of the invasive component was < or = 5 mm in 38 cases, 6-10 mm in 84 cases, 11-15 mm in 53 cases, 16-20 mm in 16 cases, > 20 mm in 15 cases. Axillary dissection was performed immediately during the initial surgical procedure in 209 patients (79%) or secondarily in 56 (21%) according to the results of intraoperative examination of surgical specimens on frozen sections. Axillary lymph node involvement was not found in DCIS, microinvasive carcinomas or invasive carcinomas < or = 5 mm in size. Among all 206 invasive breast carcinomas, lymph node involvement was found in 7.8% (16/206) of cases. There were 9/84 (10.7%) in tumors > 10 mm, 7/122 (5.8%) in tumors < or = 10 mm. Thus, it is concluded that lymph node involvement is unlikely to be found in patients with non palpable breast cancers, specially those with carcinoma in situ, microinvasive breast tumors and invasive breast cancer with less than 5 mm maximum diameter size. Axillary dissection may be avoided in these patients. However, the use of new prognostic factors of lymph node involvement may help in the definition of patient group.

Technetium-99m-sestamibi uptake in breast tumor and associated lymph nodes

The aim of this study was to measure the accumulation of 99mTc-sestamibi in breast tumors and their axillary lymph nodes in patients undergoing scintimammography.

METHODS

Eighteen patients who were scheduled for breast surgery underwent scintimammography with 740 MBq of 99mTc-sestamibi on the day before the operation. The next morning, reinjection with 370 MBq was performed. Immediately after the surgical procedure, the 99mTc activity of the tumor samples and, when available, the related lymph nodes was measured in a gamma counter. The samples were weighed and prepared for histological analysis. The activity of each sample was normalized to the mean activity of normal tissue samples obtained from the same patient.

RESULTS

Among the 198 samples analyzed, the relative uptake of sestamibi was increased in 111 containing normal lymph nodes (1.80+/-0.79 vs 1.00+/-0.22, p<0.05), as well as in the seven containing invaded lymph nodes (2.01+/-0.83, p<0.01) and more dramatically, in the 22 with a carcinoma (5.64+/-3.06, p<0.001). In two patients with a benign lesion, both scintigraphy and counting demonstrated increased activity in the tumor. Four patients had negative scan results despite the presence of malignant tumor and a more than fourfold increase of sestamibi concentration in two of them.

CONCLUSION

Technetium-99m-sestamibi concentrates strongly in breast carcinoma, sometimes even when the scan results appear normal, and mildly in lymph nodes, especially when invaded; it also concentrates in some benign tumors, possibly in relation to the presence of epithelial hyperplasia.

[Immediate breast reconstruction following mastectomy for recurrence after conservative treatment]

The authors report their experience with immediate breast reconstruction following salvage mastectomy for local recurrence of invasive breast cancer after conservative treatment. From January 1987 to December 1992, among 91 immediate breast reconstructions, 41 were performed for this indication. The results and complications were analyzed for each of the three techniques used: simple prosthesis (n = 5), latissimus dorsi flap with prosthesis (n = 26), trans abdominalis musculocutaneous flap--TRAM flap (n = 10). Immediate flap reconstruction appears to be required in cases of either early or late local relapse since the skin has been irradiated and the mastectomy must be extensive. The transrectus abdominalis flap allows more extensive resection and gives better cosmetic results without the need for a prosthesis.

Treatment results, survival and prognostic factors in 109 inflammatory breast cancers: univariate and multivariate analysis

Between 1978 and 1987, 109 patients without metastatic disease were treated by induction chemotherapy for inflammatory breast cancer (IBC) or "neglected" locally advanced breast cancer (LABC): 62 patients had a clinical history of rapidly growing tumours (doubling time < or = 4 months) and inflammatory signs; conversely, the 47 neglected patients had local inflammation with a longer history of LABC. 103 patients were fully evaluable. All patients received the same induction chemotherapy with doxorubicin, vincristine, cyclophosphamide and 5-fluorouracil. After six cycles, locoregional treatment was by radiotherapy if a complete or nearly complete response had been obtained, and total mastectomy, with pre or postoperative radiotherapy, in other cases. The chemotherapy after local treatment comprised of six cycles for LABC and 12 cycles for IBC (six without doxorubicin). With a median follow-up of 120 months, the median overall survival (OS) time was 70 months as against 45 months for disease-free survival (DFS). No difference was observed for OS and DFS between LABC and IBC. The regional recurrence rate was 24% (15% for radiotherapy alone). 20 factors of potential prognostic significance were evaluated by univariate and multivariate analysis. For DFS and OS, univariate analysis suggested a worse prognostic significance for "peau d'orange" appearance of the skin, clinical evidence of node involvement and poor response to chemotherapy after three cycles, on mammographic criteria. The cumulative dose of doxorubicin after three cycles seemed to have a significant effect on OS (P < 0.03) but was too closely correlated with age to draw definite conclusions. In the multivariate analysis, "peau d'orange", menopausal status and clinical node involvement predicted DFS. "Peau d'orange" and clinical node involvement also predicted OS. Our results indicate that IBC and LABC do not behave differently when treated with our procedure.

Neoadjuvant chemotherapy in 126 operable breast cancers

126 patients with non-inflammatory operable breast cancer, who otherwise would have undergone modified radical mastectomy (MRM), were treated by induction chemotherapy. Before treatment, every patient had a local and general assessment, and pathological or cytological evidence of malignancy. Patients received, every 3 weeks, the same treatment with doxorubicin, vincristine, cyclophosphamide, 5-fluorouracil (AVCF); methotrexate was added in 80 cases (AVCFM). Tumour shrinkage greater than 50% was documented in 105 (83%) of the 126 women. A higher objective response rate was obtained in aneuploid or high S phase tumours, especially in the patients treated with methotrexate. After chemotherapy, 41 patients were then treated by radiotherapy alone after complete or sub-complete response; 64 had a residual tumour that could be treated by conservative surgery and radiotherapy. Only 19 had MRM and radiotherapy. Histopathological complete remission was documented in 1 case; isolated residual tumour cells were found in 5 patients. Thus primary chemotherapy enhanced the possibility of breast conservation in up to 83% of the cases in a series in which most would have been otherwise subjected to a MRM because of tumour size.

Clinical relevance of immunohistochemical detection of multidrug resistance P-glycoprotein in breast carcinoma

In 20 women with breast carcinoma, 17 of whom had locally advanced cancer and 3 of whom had confirmed metastases, the expression of P-glycoprotein was evaluated before the start of a chemotherapy regimen that included multidrug resistance-related drugs. With the use of the C494 monoclonal antibody in an avidin-biotin-immunoperoxidase technique, P-glycoprotein was detected in 17 of 20 tumor samples. Results were expressed in a semiquantitative manner, taking into account the number of positive tumor cells (N index) and the specific staining intensity (I index). The 17 patients with nonmetastatic cancer were followed from the first cycle of chemotherapy to cancer recurrence; subsequent to six cycles of chemotherapy, all of these patients except one were rendered clinically disease-free through surgery and/or radiation. The end point was defined as either local/regional recurrence or metastasis. Strong P-glycoprotein-positive staining in a majority of tumor cells (the N+/I+ phenotype) was significantly correlated with no initial response to chemotherapy (P less than .02) and with a shorter progression-free survival (P less than .02). Thus, the pretreatment evaluation of P-glycoprotein expression may be of prognostic value in patients with locally advanced breast cancer.

[Surgical technic for the excision of subclinical radiologic anomalies of the breast]

Routine mammography screening can detect radiologically evident but subclinical breast anomalies. Insofar as these anomalies can correspond to carcinoma, either in situ or invasive, surgeons are increasingly forced to operate on non-palpable breast lesions. A regulated and strict surgical technic is necessary to ensure exeresis of a radiologically detected focus, while avoiding sufficient adjacent healthy parenchyma and thus reducing the esthetic prejudice to a minimum. Technical artifices are described to assist compliance with these imperatives. First, the periareolar incision is chosen, as a principle, for its cosmetic qualities. Insofar as it provides an often limited approach it can be extended radially. The latter is concealed during closure without major alteration to the areola. Second, the radiating exeresis assists detection and is safer. It also provides glandular reconstruction of good esthetic quality and facilitates follow up surveillance.