SEARCH

SEARCH BY CITATION

Keywords:

  • quantitative immunocytochemistry;
  • molecular signature;
  • breast cancer;
  • prognosis

Abstract

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Quantitative immunocytochemical assays of 1,200 breast carcinomas were assessed after construction of tissue microarrays. A total of 42 markers were evaluated for prognostic significance by univariate log rank test (mean follow-up, 79 months), using quantitative scoring by an image analysis device and specific software. Complete data were obtained for 924 patients, for whom 27 of the 42 markers proved to be significant prognostic indicators. Analysis of these 27 markers by logistic regression showed that 18 (cMet, CD44v6, FAK, moesin, caveolin, c-Kit, CK14, CD10, P21, P27, pMAPK, pSTAT3, STAT1, SHARP2, FYN, ER, PgR and c-erb B2), and 15 when ER, PgR and c-erb B2 were excluded, were 80.52% and 78.9% predictive of disease outcome, respectively. The immunocytochemical assays on 4 micron thick sections of fixed tissue are easy to handle in current practice and are cost-effective. Quantitative densitometric measurement of immunoprecipitates by computer-assisted devices from digitized microscopic images allows standardized high-throughput “in situ” molecular profiling within tumors. It is concluded that this 15 marker immunohistochemical signature is suitable for current practice, since performed on paraffin sections of fixed tumor samples, and can be used to select patients needing more aggressive therapy, since this signature is about 80% predictive of poor clinical outcome. Also, the markers included in the signature may be indicative of tumor responsiveness to current chemotherapy or suggest new targets for specific therapies. © 2008 Wiley-Liss, Inc.

Several studies over the last decade have reported genomic and transcriptional abnormalities in breast carcinomas.1–9 Recently, molecular signatures predictive of prognosis have been documented and recommended for the management of patients with breast carcinomas.3, 5–8 The procedures for identification of such signatures in individual tumors require frozen tissue fragments obtained by trained pathologists using fast appropriate sampling, or frozen tissue prior to fixation in specific fixatives. Moreover, DNA microarrays require expensive laboratory supplies. The resulting high cost of such procedures and the time required for tissue sampling and DNA assay processing make it difficult to recommend these tests for use in everyday current practice at the time of histological diagnosis in individual patients.

In contrast, recent developments in immunohistochemistry show that immunodetection of posttranscriptional protein products of some of the reported prognostic indicator genes within tumor tissue is economically relevant. Such procedures may be suitable for routine practice, since they involve much lower costs, are applicable to formalin-fixed and paraffin-embedded tissue fragments and use documented and commercially available antibodies. Also, they require small amounts of tissue (4 μm per section and per antibody or protein or marker tested) that are easily obtained from remaining paraffin blocks after microscopic examination.

In this respect, recent studies have reported immunohistochemical profiles of breast carcinomas of various types according to the new taxonomic classification based on DNA array profiling, including luminal A and B and Her-2, normal and basal-like carcinoma subtypes.10–23 The latter subtype lacks tailored therapies such as hormone or anti–c-erb B2 therapies.

The main concern in routine practice for early diagnosis of breast carcinomas, also with a close relationship to research aims, is the identification within tumors of molecules which can be potentially blocked by new therapies that specifically target these molecules. In addition, markers of disease outcome are needed, to direct more aggressive treatment specifically to patients with a molecular profile associated with poor prognosis among those categorized in the same subgroup by clinicopathological criteria (namely small node-negative grade 2 tumors).

In view of the high-cost of screening, diagnosis and therapy of breast carcinomas, a simplified cost-effective means of identifying in situ protein signatures, detectable by immunohistochemistry and indicative of poor outcome in patients who then need more aggressive or specific therapies, appears to be a relevant alternative to genomic assays that are more appropriate to basic and academic purposes.

However, the immunocytochemical procedures must be standardized as far as possible before they can be recommended for clinical practice. In particular, quantification of immunoprecipitates with automated computer-assisted devices relying on densitometry allows more objective analysis of results.24–31 Also, quantitative data are more appropriate for statistical analysis and permit more valid studies, although some variation in interpretation of results can still occur.32

Our objective in the present study was to determine immunohistochemical criteria for phenotyping of tumors of poor prognosis, and particularly for metastatic risk that would be economically applicable for individual patients with breast carcinoma, using the latest methodology for standardization and quality control. We have focused on quantitative densitometry of immunoprecipitates on digitized microscopic images to provide accurate numerical data that are compatible with requirements for modern statistical analysis (log rank, logistic regression, unsupervised hierarchical clustering). More precisely, in this study we aimed first to confirm the prognostic value of reported immunocytochemical markers, when evaluated with high-throughput standardized assays using tissue microarrays (TMAs)12–23, 31 in a large retrospective series of breast carcinomas (n = 924); second, to quantify the immunohistochemical precipitates within digitized microscopic TMA images, using an automated computer-assisted device; and third, to correlate the quantified immunohistochemical expression of each marker and of groups of markers with patients' outcome. The overall goal was to identify the best group of markers, in terms of sensitivity and specificity, to predict prognosis within 48 hr on tissue sections that would be suitable in clinical practice for individual patients at the time of diagnosis, simultaneously with pathological reporting.

The selected markers included a set of 42 prognostic markers known to be indicators of tumor cell growth, proliferation and scattering and of angiogenesis, in addition to markers of some signaling pathways (see review22).

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Patients

The subjects were a consecutive series of 1,200 patients with invasive breast carcinomas who were operated on from 1995 to 2002 (mean follow-up, 79 months) in the same department at the Hôpital Conception, Marseille. Surgery was in all cases the first treatment (PB). For this first step of treatment, patient management was handled by the same group of surgeons and by 3 senior pathologists (CC, SG and LA). Conservative treatment, mastectomy and node resection (complete or sentinel) were applied according to current European recommendations. Likewise, radiotherapy, chemotherapy and hormone therapy were applied according to criteria currently used at that time.

Analysis of the distribution of the series by age, histological type and grade, and nodal status before TMA construction revealed the usual distribution of breast carcinomas and no bias in tumor selection as compared with literature data. Because of technical difficulties in performing immunocytochemical tests on many serial paraffin sections of a TMA to evaluate the 42 different markers, complete data for all markers were finally obtained for only 924 patients of the initial series of 1,200.

The 2005 follow-up data in clinical records showed that 181 of 924 were metastatic, and 32 patients deceased.

Our study focused mainly on correlation of quantitative immunohistochemical data with patients' outcome. Current histoprognostic criteria on H&E staining were not retained for statistical analysis, mainly to limit the burden of data and also to focus the statistical analysis on continuous variables homogeneously obtained by (numerical) densitometric measurements of immunoprecipitates with the image analyzer.

Tissues

Tissue samples were all taken from surgical specimens after formalin fixation. Attention was paid to optimal consistent tissue-handling procedures, including fast immersion in buffered formalin in an appropriate container by pathologists or by nurses trained in the procedure. Tumor fragments were large and thick enough to allow further TMA construction. Duration of fixation was 24 hr for smaller samples (<5 cm) and 48 hr for larger ones, to improve formalin penetration, before specimen dissection at room temperature. After fixation, paraffin pre-embedding and embedding were performed in currently available automated devices of the same brand.

Paraffin blocks were stored in the same room, where temperature was maintained at 20°C prior to TMA construction.

TMA construction

The procedure for construction of TMAs was as previously described.22, 31 Briefly, cores were punched from the selected 1,200 paraffin blocks (from 1,200 patients), distributed in 6 new blocks including 2 cores for each tumor (200 cases per block, a total of 2,400 cores) of 0.6 mm diameter. To avoid false positive staining that might result from stromal inflammatory cells that could react with the antibodies tested, the tumor areas selected for the TMA 0.6 mm punch were dense carcinomatous areas with minimal stroma including essential vessels and few fibroblasts. For those tumors with low-epithelial structure density and wide stromal component, areas lacking inflammatory reaction, if any, were selected. Also, inflammatory carcinomas were excluded of this series and are specifically under investigation in other TMA (work in preparation).

All the new blocks (TMAs) were stored at 4°C, before that sections (4 μm thick) were prepared for each marker to be examined by immunohistochemistry.

Immunohistochemistry

Serial tissue sections were prepared and stored at 4°C, 24 hr before immunohistochemical processing, as previously reported.22, 31 Immunoperoxidase procedure was performed using an automated Ventana Benchmark XT device and Ventana kits.

Markers were detected using commercially documented antibodies (Table I). Dilutions of antibodies were determined by a prescreening on the usual full 4 μm thick sections prior to use on TMA sections.

Table I. Antibodies used in the Study
 AntibodiesSupplierSourceClone
  1. Mmab, mouse monoclonal antibody; Rmab, rabbib monoclonal antibody; Rpab, rabbit polyclonal antibody.

1ERVentanaMmab6F11
2PgRVentanaMmab1E2
3c-erbB2NovocastraMmabCB11
4P16NeomarkersMmabAb7(16PO7)
5P53DakoMmabDO-7
6Bcl2DakoMmab124
7CD 146NovocastraMmabN1238
8CD 105DakoMmabSN6h
9Caveolin 1Santa CruzRpab 
10cMetChemicon/AbcysMmab4AT44
11JAK1Cell SignalingRpab 
12PI3 kinaseCell SignalingRpab 
13PTENCell SignalingMmab26H9
14Cytokeratins 5-6DakoMmabD5/16B4
15CD 117 (c-Kit)DakoRpab 
16E-CadherinZymedMmab4A2C7
17CA IXAbcamRpab 
18Cytokeratin 903DakoMmab34BE12
19P63DakoMmab4A4
20FYNAbcamMmab1S
21SHARP 2AbcamRpab 
22P21Waf1-Cip1Cell SignalingMmabDCS60
23P27 Kip1Cell SignalingRpab 
24P38 MAP kinaseCell SignalingRpab 
25FAKCell SignalingRpab 
26STAT-1Cell SignalingMmab9H2
27EGFRVentanaMmab3C6
28Phospho-MAPKAPK-2Cell SignalingRmab(Thr334)
29Cytokeratin 19DakoMmabBA17
30VimentinImmunotechMmabV9
31CD34DakoMmabQBEnd-10
32CD10NovocastraMmab56C6
33STAT-3Cell SignalingRmabTyr 705 D3A7
34Cytokeratin 14NovocastraMmabLL002
35Cytokeratin 17DakoMmabE3
36Cytokeratins 8 and 18ZymedMmabZym5,2(UCD/PR-10,11
37Moesin 1BiomedaMmab38/87
38CD44v6NovocastraMmabVFF-7
39Ezrin(p81,80k,cytovillin)NeomarkersMmab3C12
40P-CadherinNovocastraMmab56C1
41MaspinBD PharmingenMmabG167-70
42FGFR-1 Flg (C-15)Santa CruzRpab 

Specificity of signaling molecules was documented by the suppliers. Those antibodies recognizing phosphorylated molecules are identified with “p” sign such as pSTAT3. In contrast, those simply indicated like STAT1, FYN, focal adhesion kinase (FAK), recognize nonphosphorylated molecules.

Image analysis

Automated densitometric measurements of immunoprecipitates in cores were assessed for each marker antibody in each core individually identified after digitization and image cropping of the slides, as previously reported.22, 31 Briefly, TMA analysis with a SAMBA 2050 automated device (SAMBA / TRIBVN, Châtillon 92320, France)24–27 was performed according to the following protocol.

First, an image of the entire slide was built up using low-power magnification (2×). This image was made up of a mosaic of images acquired along a rectangular grid with contiguous fields. Second, the area of the slide containing the TMA cores was automatically delineated and scanned at higher magnification (20×, pixel dimension 7.4 μm). Third, after autofocusing, the images were acquired with an overlap greater than the largest mechanical positioning error. Using the image contents, a matching algorithm determined precisely the relative position of each image with respect to its neighbors. Calculated overlap was removed from images to produce a new set of higher-magnification images, thus covering precisely the cores of interest. A specially developed tool referred to as TMA crop then allowed superposition of the TMA grid onto the reduced image and precise alignment of each node of the grid with the core location within the image. The final step was performed automatically using the core image contents to ensure pixel precision of the match. From the images acquired with 20× magnification, a new set of images was next computed, one for each core. For color analysis of the core images, the SAMBA “immuno” software was applied as previously reported24–31 in usual full tissue sections.

In this study, we correlated the patients' follow-up parameters with a quantitative score combining the surface stained and the intensity of staining22, 31 computed by the SAMBA “immuno” software.

Statistical analysis

Immunohistochemical expression of each marker was first correlated with patients' disease-free survival using NCSS and Statistica statistical software.

When significant differences in mean expression were identified in patients with disease and without disease, the prognostic significance was determined by log rank tests (Kaplan-Meier curves). The appropriate threshold of prognostic significance for a given marker was determined as previously recommended33 and described.22, 24–31

Logistic regression (with ROC curves) was then used to identify the combination of markers with the best sensitivity and specificity indicative of a proteomic signature of poor prognosis.

Finally, unsupervised hierarchical clustering of significant prognostic indicators in the overall series provided qualitative data to be compared with previously reported research results on the relationship and on the role played by these molecules in the process of cancer metastasis.

Results

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Distribution of positive staining

Whatever the positive immunostaining location (nuclear, cytoplasmic or cell membranes), the microscopic images enclosing all spots in individual slide were digitized. After “cropping” of the images, densitometry was assessed by the analyzer in each spot identified by cropping with the SAMBA software specifically developed for immunohistochemically stained sections. The degree of the immunostaining was automatically evaluated and quantitative scores were consequently computed by the software.

Screening of potential markers of prognosis

The 42 markers tested (Tables I and II) were selected on the basis of literature data on breast carcinoma prognostic markers and our experience of immunostaining quality in pretests of commercially available antibodies on frozen tissue and current full paraffin sections, prior to high-throughput immunodetection on TMAs including the series of 1,200 tumors (Fig. 1).

thumbnail image

Figure 1. (ad) TMA enclosing 268 cores (0.6 microns in diameter) from breast carcinomas. An example of immunohistochemical test with anti CD44v6 and “cropping” procedure consisting in superimposing a standard grid on the TMA image by a modifying grid shape in order to separate each spot prior to densitometry measurement. Higher magnification of spots immunostained with specific anti (e) -CD44v6, (f) -cMet, (h) -pMAPKinase and (g) -FAK. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Table II. Correlation of Quantitative Immunocytochemical Score of 42 Markers (Image Analysis) with Disease-Free Survival (Mean Follow-Up 79 Months) in 924 Breast Cancers
Tissue markers (immunohistochemistry): Positive tumorsImage analysis (densitometry): Quantitative score of positive immunostaining
Mann-WhitneyChi-square test
pMean value– disease-freeMean value– with diseasepPatient number disease-freePatient number with disease
Tumor progression invasion/cell adhesion
1CMet485/924<0.00112.326.9<0.0145/743120/181
2CD44v6249/9240.00012.631.8<0.0001102/743147/181
3FAK217/9240.00011.53.10.001180/743137/181
4Moesin78/9240.00112.527.70.000217/74361/181
5Ezrin792/924<0.00011.95.1<0.000112/74380/181
6E-cadherin791/924<0.00834.917.6NS  
7P-cadherin104/924<0.000136.616.9<0.00198/74378/181
Angiogenesis
8CD105112/924<0.0011.75.3<0.0130/74382/181
9CD146710/9240.0210.92.80.026629/743181/181
10CD34737/9240.032.95.8NS  
11CD31632/924NS  NS  
12CAIX121/924<0.00012.37.1<0.0019/743112/181
Tumor growth
13c-erb B281/924<0.000112.432.6<0.00014/74177/181
14EGFR121/924<0.0017.913.10.00641/74191/181
15C-Kit221/924<0.00013.58.9<0.001109/741112/181
16FGFR144/9240.00523.238.60.00759/74365/181
Cell proliferation
17P 53625/924<0.00011.425.2<0.0001444/743181/181
18P 16371/924<0.017.116.4NS  
19P21321/924<0.0013.911.8<0.001170/743151/181
20PTEN653/9240.0016.42.50.018531/74322/181
21Maspin597/924<0.0018.916.4<0.0001161/743135/181
22Bcl2140/924<0.012.811.30.0452/74378/181
23P27284/924<0.013.77.20.003207/74357/181
24P6357/924<0.011.74.10.00725/74332/181
Signaling
25P38 MAPK178/9240.00142.34.50.000185/74393/181
26P13K457/924<0.00115.728.90.004305/743152/181
27pMAPK393/923<0.00112.717.6<0.01271/743112/181
28pSTAT-3312/9230.00633.16.4<0.01260/743152/181
29STAT-1198/923<0.00014.59.7<0.00137/743161/181
30FYN669/923<0.019.314.9<0.00155/743114/181
31JAK219/9240.00361.54.40.0023117/743142/181
32SHARP-2837/9240.046.58.8<0.001680/743157/187
Cell subtypes
33CK5.6207/924<0.00013.47.10.003125/74382/181
34CK14234/924<0.011.36.9<0.001107/743127/181
35CK17230/9240.031.68.20.0074128/743112/181
36CK8.18624/9240.002317.411.5NS  
37CK19601/924<0.00015.30.6<0.0001596/74325/181
38CD1061/9240.00461.12.10.00119/74342/181
39Vimentin703/924<0.0012.16.4NS  
40Caveolin-1137/924<0.00001924.3<0.0000165/743172/181
41ER761/924<0.0000142.38.10.001604/74358/181
42PgR503/924<0.0000136.16.20.0001411/74392/181

In the first step of the study, immunoexpression of markers was screened in dedicated TMAs containing tumors from disease-free patients and from patients with metastasis and recurrent disease. The first step in assessing prognostic value consisted in comparison of mean quantitative scores in relation to the number of positive and negative patients in the disease-free and diseased categories (Mann-Whitney and chi-squared tests) (Table II). The positive or negative correlation of the markers expression is indicated in Table II. Quantitative scores in patients with disease-free survival greater than those with metastases, indicate a positive correlation.

Conversely quantitative scores in patients with disease-free survival, smaller than in patients with metastases indicate a negative correlation.

Data for all markers were obtained for only 924 out of the 1,200 patients because of the loss of cores after immunostaining procedures for some antibodies, and tumors lacking data for all 42 markers were not further considered.

Quantitative score and survival

The prognostic significance of markers was further individually evaluated by a univariate log rank test (Kaplan-Meier survival curves).

The threshold of positive staining was first established, determined by the image analysis device as 0 or >0 computed quantitative scores. In positive (>0) cases, optimal thresholds greater than zero, specific for each marker, were determined according to the p value curves from univariate analysis (log rank), as reported by Altman et al.33 and previously used,22, 24–31 as shown in Figure 2 and Table III.

thumbnail image

Figure 2. Kaplan-Meier curves (log rank test) and p value curves (according to Altman et al.33) used to determine the threshold of quantitative scores for prognostic significance of markers; in this example, cMet, P21, maspin, FAK and in 924 breast carcinomas (TMA, quantitative immunochemical assays).

Download figure to PowerPoint

Table III. Optimal thresholds of quantitative scores for 27 markers of prognostic significance in univariate analysis log rank test, as determined according to the method of Altman et al33 for 924 patients with breast cancer
  p (log rank)Quantitative score thresholdImmunostaining localization
  1. Cyto, cytoplasm; Mb, cell membrane.

1Caveolin<0.0129Cyto and mb
2CD10<0.014Mb
3CD146<0.0012.3Cyto and mb
4CD 44 v6<0.00111.9Mb
5c-erb B2<0.00128Mb
6CK19<0.014.6Mb
7EGFR<0.019.4Mb
8cMet 2<0.0117Nuclear
9cKit<0.00112.9Mb and cyto
10Ezrin<0.0013.9Mb
11FAK<0.0012Cyto
12FGF-R<0.0123Cyto and mb
13FYN<0.012.7Cyto
14CK14<0.0018.3Cyto
15Maspin<0.0015.7Cyto and nuclear
16Moesin<0.000116.4Cyto and mb
17P16<0.017.5Nuclear and cyto
18P21<0.0012.1Cyto and nuclear
19P27<0.015.3Nuclear
20P38<0.0011.1Cyto and nuclear
21pMAPK<0.00112.5Nuclear
22pSTAT3<0.014.4Nuclear
23PTEN<0.00012.7Cyto
24STAT1<0.000110.7Nuclear
25ER<0.019.3Nuclear
26PgR<0.00017.9Nuclear
27SHARP 2<0.0015.6Nuclear

Table III shows that 27 of the 42 prognostic markers were significant in the univariate analysis and Figure 3 illustrates an unsupervised hierarchical clustering of prognostic significant markers in the log rank test.

thumbnail image

Figure 3. Unsupervised hierarchical clustering of the 27 markers with prognostic significance in the log rank test, from quantitative densitometry of immunohistochemical assays on TMA (n = 924 patients with breast carcinoma). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Logistic regression (ROC curves)

To determine a proteomic immunocytochemical signature of poor outcome, those markers of prognostic significance in univariate log rank tests were reevaluated by logistic regression (Table IV), which showed that 18 of the 27 markers remained significant predictors of prognosis after a first regression step (p < 0.05). Importantly, with the 27-marker signature, 82.14% of the patients were well classified in either the good or poor prognostic category, with a sensitivity of 85% and specificity of 80.6% (Fig. 4).

thumbnail image

Figure 4. (a and b) First step of logistic regression and ROC curves determining the signature with 27 (a) and with 18 (b) (out of 27) markers that well classified 82.14% (a) and 80.52% (b) the patients in the category of poor prognosis or (c and d) second step of logistic regression with 24 (c) and 15 (d) markers (without ER, PgR and c-erb B2) that well classified patients in 80.73% (c) and 78.9% (d), respectively. Quantitative immunohistochemical assays on TMA (n = 924 breast carcinomas).

Download figure to PowerPoint

Table IV. Logistic Regression of 27 Markers with Prognostic Significance in Univariate Log Rank Test (TMA/924 Breast Carcinomas)
 Immunocytochemical markerp (regression)
1Cav0.00064
2CD100.00008
3CD1460.37793
4CD44v60.00352
5C-erb B20.00739
6CK190.075
7C-KIT0.00902
8CMet0.00034
9EGFR0.30586
10Ezrin0.53581
11FAK0.03594
12FGF-R0.85278
13FYN0.00936
14CK140.03244
15Maspin0.1586
16MoesN0.04428
17P160.17401
18P210.00177
19P270.0032
20P380.14782
21pMAPK0.0107
22pSTAT-30.0012
23PTEN0.6851
24ER0.00443
25PgR0.0027
26SHARP20.00159
27STAT10.0021

When the signature logistic regression was assessed with exclusion of ER, PgR and cerbB2 (24 markers instead of 27), the sensitivity and specificity remained similar (82.32% and 81.76%, respectively), with 80.73% patients well classified (Table V and Fig. 4). This 15-marker signature includes cMet, CD14v6, FAK, moesin, caveolin, c-Kit, CK14, CD10, P21, P27, pMAPK, pSTAT3, STAT1, SHARP2 and FYN. Markers related to tumor cell motility and spreading were of special interest, since our study aimed at prediction of patients' outcome that is closely linked to development of metastases. Likewise it is not surprising to observe that molecules such as FAK, Erk/PAK-P21, MAPK/P38, STAT1, STAT3 of cMet activation were also found in our signature, like FYN involved in signaling pathways of angiogenesis and skeleton rearrangement.

Table V. Logistic Regression for the 24 Markers with Prognostic Significance in the Univariate Log Rank Test (27 Markers of Table IV, Excluding ER, Pr and c-erb B2)
 Immunocytochemical markerp
  1. Fifteen markers are retained as prognostic indicators.

1Cav0.00051
2CD100.00016
3CD1460.26176
4CD44v60.00021
5CK190.36523
6c-KIT0.02624
7cMet0.0012
8EGFR0.57889
9Ezrin0.59976
10FAK0.00172
11FGF-R0.74405
12FYN0.00209
13CK140.01179
14Maspin0.19778
15Moes0.03426
16P160.21716
17P210.00521
18P270.0045
19P380.39845
20pMAPK0.02948
21pSTAT30.00002
22PTEN0.59787
23SHARP20.0036
24STAT10.0026

When a second step of regression was performed with either the 18 or 15 markers, the percentages of well classified patients were very close (80.52% and 78.91%) (Tables VI and VII, Fig. 4) and 17 or 14 markers remained significant (Table VI).

Table VI. Second Logistic Regression from Table V (15 Markers Retained in the First-Step Regression)
 Immunocytochemical markerp
1Cav0.00018
2CD100.00023
3CD44v60.00098
4c-KIT0.00873
5cMet<0.0001
6FAK0.00121
7FYN0.00032
8CK140.01747
9P160.04533
10P210.00088
11P27<0.0001
12pMAPK0.06893
13pSTAT30.00001
14SHARP20.00008
15STAT1<0.0001
Table VII. Logistic Regression with and without ER, PR and c-erb B2: Signature of 18 or 15 Prognostic Markers
Immunocytochemical markersSensitivity (%)Specificity (%)Well classified (%)Patients well classified (n = 924) with positive signature
DiseaseDisease-free
All markers
1st step of regression (n = 27 markers)8580.682.1427/181144/743
2nd step of regression (n = 18 markers)80.3582.380.52149/181146/743
Without ER, PgR, c-erb B2
1st step of regression (n = 24 markers)82.3281.7680.73148/181147/743
2nd step of regression (n = 15 markers)81.7680.278.9149/181149/743

Discussion

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Using our quantitative immunohistochemical procedure, we first determined the cut-off points for immunocytochemical expression of each marker having prognostic significance in terms of disease-free survival in a univariate log rank test, in our series of 924 breast carcinomas (Table II). We next determined, by logistic regression, the best association of prognostic indicators (Tables IV and V) and showed that 78.9% prognostic prediction is provided by a 15-marker signature that includes cMet, CD44v6, caveolin, FAK, moesin, c-Kit, CK14, CD10, P21, P27, pMAPK, pSTAT-3, SHARP-2, STAT-1, FYN (Table VI) independently of ER, PgR and c-erb B2 status.

No previously reported study has used this quantitative immunocytochemical approach to tumor profiling in terms of predicting prognosis in breast carcinoma patients. We used an immunohistochemical procedure that is much easier to handle than genomic profiling in routine clinical and pathological practice, requiring very few tissue samples, since it can be applied on tissue remaining within paraffin blocks after diagnosis, is performable in 24 hr in pathological laboratories and costs about 20-fold less than commercially available genomic tests.

The selection of markers was based on a literature review covering principal markers involved in tumor growth and progression, and on our previous experience in their immunocytochemical expression in tissue sections from frozen or fixed tissues and TMAs.24–31

Recent studies have underlined the role of cMet in tumor spreading (see review22).

We and others have shown the relationship with poor prognosis in breast cancers of this marker and also of CD44v6 (see review30). Several recent studies have shown the synergistic role of CD44v6 and cMet in tumor cells of several types34 and the interaction of the EMR (ezrin-moesin-radixin) superfamily and CD44v6 for HGF activation of cMet to promote the ERK signaling cascade, inducing cell migration. Moreover, EMR components act on cell adhesion (integrin β2) and the cytoskeleton.35, 36 FAK is also known to play a pivotal role in the control of integrin-mediated cell functions including cell migration, progression and survival, coacting with cMet and EMR.37 Consistent with these findings, poor outcome and high metastatic risks associated with CD44v6, cMet, EMR and FAK immunoexpression in breast carcinoma have been individually documented in several previous studies.17, 22, 30, 31

It is therefore not surprising that in the present study, overexpression of CD44v6, cMet, moesin and FAK was found to be included in the immunocytochemical signature of poor prognosis as determined by logistic regression for early (79 months mean follow-up of patients in our study) metastatic disease.

Caveolins are membrane proteins involved in membrane trafficking, gene regulation, signal transduction and mediation of intracellular processes, as well as in carcinogenesis, being over-expressed in invasive breast carcinomas.38, 39 Conflicting results on the prognostic significance of caveolin expression in breast carcinomas have been reported.39 More specifically, caveolin 1 has been reported in basal-like and metaplastic breast carcinomas15, 17, 39.Interestingly, our results show that caveolin 1 expression in invasive breast carcinomas is definitely associated with poor prognosis, both in univariate analysis and after logistic regression along with markers reflecting increased motility of tumor cells.

Overexpression of molecules with prognostic significance in cancer results from amplification or mutation of genes, or from epigenetic processes including decreased methylation or increased acetylation. When overexpression is observed, the main interest is to evaluate whether the proteins are activated and involved in cell transduction processes. The results from our first-step regression suggest that transduction pathways are activated with prognosis-significant expression of major signaling proteins such as PI3K, pMAPK, pSTAT3, in addition to more specific signaling pathways, such as SHARP-2, FYN, STAT-1 and FAK (see review22, 31). This may imply that the proteins shown to be expressed are activated. However, only pSTAT-3, STAT-1 and SHARP-2 were still significantly correlated with poor prognosis in the reduced signature (15 markers) after the second step of logistic regression.

Increased tumor vascularization and angiogenesis are associated with poor outcome of patients (see reviews22, 31, 41). We analyzed several markers, such as CD105, Tie2, CD34, VEGF-R1 and -R2, CD146,40–44 in frozen tissue to evaluate angiogenesis in breast cancers associated with poor prognosis. In this study on fixed tissue, only CD146 proved to be of prognostic significance when evaluated on TMAs in univariate analysis, and not in logistic regression.

New approaches to molecular typing of tumors are relevant to prognosis, but also to prediction of response to therapy. Therefore, markers identified as prognostic indicators can also be regarded as indicators of responsiveness to current chemotherapies (like P21) or as targets for tailored therapies. For example, caveolins, moesin and CD44v6 have been shown to be indicators of responsiveness to anthracyclines and paxitaxel.45, 46 Also, specific therapy with agents such as dasatinib (a small molecule orally active as a kinase inhibitor and biologically active in cell lines with elevated caveolins, moesin and yes associated protein 1 expression) can be efficient in breast carcinomas expressing these molecules.47 Dasatinib is also active against PDGF-R and c-Kit and has been shown to be effective in leukemia after failure of imatinib therapy.48 Likewise, sunitinib, that is recommended in gastrointestinal stromal tumors (GIST) expressing c-Kit and in advanced kidney carcinomas, is an inhibitor of angiogenic tyrosine kinase receptors such as PDGR, VEGFR, FLT3 and c-Kit. This suggests that c-Kit in breast carcinomas of poor prognosis could also be targeted by sunitinib. However, response to imatinib and dasatinib or sunitinib in breast carcinomas with c-Kit expression remains to be demonstrated.

Antibodies against cMet and small molecules such as PHA66752 or kerin that target cMet, or the NK4 molecule that blocks HGF binding to cMet have been reported to act as specific cMet inhibitors in breast cancer,22 whereas other tyrosine kinase inhibitors such as Iressa, Tarceva, Herceptin and Genifinib do not inhibit cMet activity (see review49) and breast carcinomas expressing cMet may be responsive to such a tailored therapy.

Activation of cMet results in tumor cell mobility, dissociation, invasion and adhesion to the extracellular matrix via known signaling pathways37 involving PI3K in addition to FAK and ERK/PAK-P21, MAP kinase, STAT-1 and STAT-3, that are responsible for branching morphogenesis (see review32). All these are included in our signature. Small molecules have been reported to act directly against these pivotal kinases and also against Gab-1. In particular, PHA66572 has been reported to block PI3K function in cell lines from small-cell and gastric carcinomas and gliomas, whereas SU11274 inhibits FAK that is responsible for loss of intracellular junctions and increased cell matrix adhesion during mobility and scatter responses in cell culture.49 So cMet downstream transducers and signaling pathways provide a range of potential specific targets.

The CD146 extracellular domain is involved in endothelial–endothelial cell adhesion through tight junctions. The intracellular domain promotes the recruitment of the Src family kinase FYN as well as tyrosine phosphorylation of several intracellular proteins, including FAK (and paxillin), that are present in focal adhesion plaques.37, 50, 51 We observed increased FAK and FYN immunostaining in breast carcinomas of poor outcome compared with tumors from patients with longer survival. This is probably a result of increased cellular signals from cMet and CD146, which both act through FYN signaling pathways upon angiogenesis and rearrangement of the actin skeleton.49–51 Experimental studies have shown that anti-CD146 monoclonal antibodies inhibit proliferation and migration of endothelial cells and angiogenesis, reflected by a reduction in blood vessel density associated with tumor growth inhibition.51 This treatment exhibited no cytotoxic effects in animals, and its efficacy increased when the anti-CD146 monoclonal antibody AA98 was combined with other anticancer agents.51 Our data suggest that inhibition of CD146 and cMet in tumors overexpressing both markers should have a synergistic effect in potentially reducing angiogenesis and cell spreading, but clinical studies are required to demonstrate the efficacy of this strategy for human therapy.

In conclusion, we have used a standardized immunocytochemical procedure that is easy to perform in current practice with fixed tissue, to detect 42 potential prognostic markers. High-throughput quantitative densitometry was then applied to digitized TMA microscopic images. We were thus able to identify a simplified 15-marker signature predictive of disease outcome in 78.9% of a series of 924 patients, whatever their ER, PgR and c-erb B2 status, that could allow selection of those who can benefit from more aggressive therapies. Moreover, the markers identified may also be predictive of response to specific therapies like anthracycline/paxitaxel treatment or to more tailored therapies.

A deeper insight of the immunoprofile of the tumor subgroups included in this 924 series is now under investigation, particularly of those from node negative patients, and of “triple negative” tumor subset, using (i) additional markers and (ii) longer follow-up (up-dating with 2-year longer follow-up), but (iii) the same procedures for immunostaining automated quantification on TMA, and the same methods for statistical analysis. Our goal is now to identify an immunocytochemical signature predictive of poor outcome that would enable to select node negative patients who might benefit from more aggressive therapy and also that could significantly reduce unnecessary treatment in about 70% of node negative patients along with reduced drawbacks and costs of early breast cancer management. Our purpose is also to identify relevant new targets for specific therapy.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

We are grateful to ROCHE for supporting Master's and PhD projects (V. Secq and S. Giusiano).

References

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  • 1
    Perou CM,Sorlie T,Eisen MB,Van de Rijn M,Jeffrey SS,Rees CA,Pollack JR,Ross DT,Johnsen H,Akslen LA,Fluge O,Pergamenschikov A, et al. Molecular portraits of human breast tumors. Nature 2000; 406: 74752.
  • 2
    van de Vijver MJ,He YD,van't Veer LJ,Dai H,Hart AA,Voskuil DW,Schreiber GJ,Peterse JL,Roberts C,Marton MJ,Parrish M,Atsma D, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med 2002; 347: 19992009.
  • 3
    van 't Veer LJ,Dai H,van de Vijver MJ,He YD,Hart AA,Mao M,Peterse HL,van der Kooy K,Marton MJ,Witteveen AT,Schreiber GJ,Kerkhoven RM, et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature 2002; 415: 5306.
  • 4
    Sorlie T,Tibshirani R,Parker J,Hastie T,Marron JS,Nobel A,Deng S,Johnsen H,Pesich R,Geisler S,Demeter J,Perou CM, et al. Repeated observation of breast tumor subtypes in independent gene expression data sets. Proc Natl Acad Sci USA 2003; 100: 841823.
  • 5
    Bertucci F,Finetti P,Rougemont J,Charafe-Jauffret E,Nasser V,Loriod B,Camerlo J,Tagett R,Tarpin C,Houvenaeghel G,Nguyen C,Maraninchi D, et al. Gene expression profiling for molecular characterization of inflammatory breast cancer and prediction of response to chemotherapy. Cancer Res 2004; 64: 855865.
  • 6
    Pittman J,Huang E,Dressman H,Horng CF,Cheng SH,Tsou MH,Chen CM,Bild A,Iversen ES,Huang AT,Nevins JR,West M. Integrated modeling of clinical and gene expression information for personalized prediction of disease outcomes. Proc Natl Acad Sci USA 2004; 101: 84316.
  • 7
    Paik S,Shak S,Tang G,Kim C,Baker J,Cronin M,Baehner FL,Walker MG,Watson D,Park T,Hiller W,Fisher ER, et al. A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Engl J Med 2004; 351: 281726.
  • 8
    Chang HY,Nuyten DS,Sneddon JB,Hastie T,Tibshirani R,Sorlie T,Dai H,He YD,van't Veer LJ,Bartelink H,van de Rijn M,Brown PO, et al. Robustness, scalability, and integration of a wound-response gene expression signature in predicting breast cancer survival. Proc Natl Acad Sci USA 2005; 102: 373843.
  • 9
    Buyse M,Loi S,van't Veer L,Viale G,Delorenzi M,Glas AM,d'Assignies MS,Bergh J,Lidereau R,Ellis P,Harris A,Bogaerts J, et al. Validation and clinical utility of a 70-gene prognostic signature for women with node-negative breast cancer. J Natl Cancer Inst 2006; 98: 118392.
  • 10
    Korsching E,Packeisen J,Agelopoulos K,Eisenacher M,Voss R,Isola J,van Diest PJ,Brandt B,Boecker W,Buerger H. Cytogenetic alterations and cytokeratin expression patterns in breast cancer: integrating a new model of breast differentiation into cytogenetic pathways of breast carcinogenesis. Lab Invest 2002; 82: 152533.
  • 11
    Makretsov NA,Huntsman DG,Nielsen TO,Yorida E,Peacock M,Cheang MC,Dunn SE,Hayes M,van de Rijn M,Bajdik C,Gilks CB. Hierarchical clustering analysis of tissue microarray immunostaining data identifies prognostically significant groups of breast carcinoma. Clin Cancer Res 2004; 10: 614351.
  • 12
    Nielsen TO,Hsu FD,Jensen K,Cheang M,Karaca G,Hu Z,Hernandez-Boussard T,Livasy C,Cowan D,Dressler L,Akslen LA,Ragaz J, et al. Immunohistochemical and clinical characterization of the basal-like subtype of invasive breast carcinoma. Clin Cancer Res 2004; 10: 536774.
  • 13
    Ginestier C,Charafe-Jauffret E,Bertucci F,Eisinger F,Geneix J,Bechlian D,Conte N,Adelaide J,Toiron Y,Nguyen C,Viens P,Mozziconacci MJ, et al. Distinct and complementary information provided by use of tissue and DNA microarrays in the study of breast tumor markers. Am J Pathol 2002; 161: 122333.
  • 14
    Jacquemier J,Ginestier C,Rougemont J,Bardou VJ,Charafe-Jauffret E,Geneix J,Adelaide J,Koki A,Houvenaeghel G,Hassoun J,Maraninchi D,Viens P, et al. Protein expression profiling identifies subclasses of breast cancer and predicts prognosis. Cancer Res 2005; 65: 76779.
  • 15
    Charafe-Jauffret E,Ginestier C,Monville F,Finetti P,Adelaide J,Cervera N,Fekairi S,Xerri L,Jacquemier J,Birnbaum D,Bertucci F. Gene expression profiling of breast cell lines identifies potential new basal markers. Oncogene 2006; 25: 227384.
  • 16
    Livasy CA,Karaca G,Nanda R,Tretiakova MS,Olopade OI,Moore DT,Perou CM. Phenotypic evaluation of the basal-like subtype of invasive breast carcinoma. Mod Pathol 2006; 19: 26471.
  • 17
    Charafe-Jauffret E,Monville F,Bertucci F,Esterni B,Ginestier C,Finetti P,Cervera N,Geneix J,Hassanein M,Rabayrol L,Sobol H,Taranger-Charpin C, et al. Moesin expression is a marker of basal breast carcinomas. Int J Cancer 2007; 121: 177985.
  • 18
    Laakso M,Tanner M,Nilsson J,Wiklund T,Erikstein B,Kellokumpu-Lehtinen P,Malmstrom P,Wilking N,Bergh J,Isola J. Basoluminal carcinoma: a new biologically and prognostically distinct entity between basal and luminal breast cancer. Clin Cancer Res 2006; 12: 418591.
  • 19
    Da Silva L,Clarke C,Lakhani SR. Demystifying basal-like breast carcinomas. J Clin Pathol 2007; 60: 132832.
  • 20
    Reis-Filho JS,Tutt AN. Triple negative tumors: a critical review. Histopathology 2008; 52: 10818.
  • 21
    Rakha EA,El-Sayed ME,Green AR,Paish EC,Lee AH,Ellis IO. Breast carcinoma with basal differentiation: a proposal for pathology definition based on basal cytokeratin expression. Histopathology 2007; 50: 4348.
  • 22
    Garcia S,Dales JP,Charafe-Jauffret E,Carpentier-Meunier S,Andrac-Meyer L,Jacquemier J,Andonian C,Lavaut MN,Allasia C,Bonnier P,Charpin C. Poor prognosis in breast carcinomas correlates with increased expression of targetable CD146 and cMet and with proteomic basal-like phenotype. Hum Pathol 2007; 38: 83041.
  • 23
    Jacquemier J,Padovani L,Rabayrol L,Lakhani SR,Penault-Llorca F,Denoux Y,Fiche M,Figueiro P,Maisongrosse V,Ledoussal V,Martinez Penuela J,Udvarhely N, et al. Typical medullary breast carcinomas have a basal/myoepithelial phenotype. J Pathol 2005; 207: 2608.
  • 24
    Charpin C,Garcia S,Bouvier C,Martini F,Andrac L,Bonnier P,Lavaut MN,Allasia C. CD31/PECAM automated and quantitative immunocytochemical assays in breast carcinomas: correlation with patient follow-up. Am J Clin Pathol 1997; 107: 53441.
  • 25
    Charpin C,Dales JP,Garcia S,Carpentier S,Djemli A,Andrac L,Lavaut MN,Allasia C,Bonnier P. Tumor neoangiogenesis by CD31 and CD105 expression evaluation in breast carcinoma tissue microarrays. Clin Cancer Res 2004; 10: 58159.
  • 26
    Charpin C,Garcia S,Bonnier P,Martini F,Andrac L,Choux R,Lavaut MN,Allasia C. Reduced E-cadherin immunohistochemical expression in node-negative breast carcinomas correlates with 10-year survival. Am J Clin Pathol 1998; 109: 4318.
  • 27
    Charpin C,Garcia S,Bonnier P,Martini F,Andrac L,Horschowski N,Lavaut MN,Allasia C. Prognostic significance of Nm23/NDPK expression in breast carcinoma, assessed on 10-year follow-up by automated and quantitative immunocytochemical assays. J Pathol 1998; 184: 4017.
  • 28
    Charpin C,Garcia S,Bonnier P,Martini F,Andrac L,Horschowski N,Lavaut MN,Allasia C. Bcl-2 automated and quantitative immunocytochemical assays in breast carcinomas: correlation with 10-year follow-up. J Clin Oncol 1998; 16: 202531.
  • 29
    Charpin C,Vielh P,Duffaud F,Devictor B,Andrac L,Lavaut MN,Allasia C,Horschowski N,Piana L. Quantitative immunocytochemical assays of P-glycoprotein in breast carcinomas: correlation to messenger RNA expression and to immunohistochemical prognostic indicators. J Natl Cancer Inst 1994; 86: 153945.
  • 30
    Charpin C,Garcia S,Bouvier C,Devictor B,Andrac L,Choux R,Lavaut MN,Allasia C. Automated and quantitative immunocytochemical assays of CD44v6 in breast carcinomas. Hum Pathol 1997; 28: 28996.
  • 31
    Garcia S,Dales JP,Charafe-Jauffret E,Carpentier-Meunier S,Andrac-Meyer L,Jacquemier J,Andonian C,Lavaut MN,Allasia C,Bonnier P,Charpin C. Overexpression of cMet and of the transducers PI3K, FAK and. JAK in breast carcinomas correlates with shorter survival and neoangiogenesis. Int J Oncol 2007; 31: 4958.
  • 32
    Michiels S,Koscielny S,Hill C. Prediction of cancer outcome with microarrays: a multiple random validation strategy. Lancet 2005; 365: 48892.
  • 33
    Altman DG,Lausen B,Sauerbrei W,Schumacher M. Dangers of using “optimal” cutpoints in the evaluation of prognostic factors. J Natl Cancer Inst 1994; 86: 82935.
  • 34
    Orian-Rousseau V,Morrison H,Matzke A,Kastilan T,Pace G,Herrlich P,Ponta H. Hepatocyte growth factor-induced Ras activation requires ERM proteins linked to both CD44v6 and F-actin. Mol Biol Cell 2007; 18: 7683.
  • 35
    Niggli V,Rossy J. Ezrin/radixin/moesin: versatile controllers of signaling molecules and of the cortical cytoskeleton. Int J Biochem Cell Biol 2008; 40: 3449.
  • 36
    Tang P,Cao C,Xu M,Zhang L. Cytoskeletal protein radixin activates integrin alpha(M)beta(2) by binding to its cytoplasmic tail. FEBS Lett 2007; 581: 11038.
  • 37
    Chen SY,Chen HC. Direct interaction of focal adhesion kinase (FAK) with Met is required for FAK to promote hepatocyte growth factor-induced cell invasion. Mol Cell Biol 2006; 26: 515567.
  • 38
    Savage K,Lambros MB,Robertson D,Jones RL,Jones C,Mackay A,James M,Hornick JL,Pereira EM,Milanezi F,Fletcher CD,Schmitt FC, et al. Caveolin 1 is overexpressed and amplified in a subset of basal-like and metaplastic breast carcinomas: a morphologic, ultrastructural, immunohistochemical, and in situ hybridization analysis. Clin Cancer Res 2007; 13: 90101.
  • 39
    Liedtke C,Kersting C,Burger H,Kiesel L,Wulfing P. Caveolin-1 expression in benign and malignant lesions of the breast. World J Surg Oncol 2007; 5: 110.
  • 40
    Garcia S,Dales JP,Jacquemier J,Charafe-Jauffret E,Birnbaum D,Andrac-Meyer L,Lavaut MN,Allasia C,Carpentier-Meunier S,Bonnier P,Charpin-Taranger C. cMet overexpression in inflammatory breast carcinomas: automated quantification on tissue microarrays. Br J Cancer 2007; 96: 32935.
  • 41
    Dales JP,Garcia S,Carpentier S,Andrac L,Ramuz O,Lavaut MN,Allasia C,Bonnier P,Taranger-Charpin C. Prediction of metastasis risk (11 year follow-up) using VEGF-R1. VEGF-R2, Tie-2/Tek and CD105 expression in breast cancer (n=905). Br J Cancer 2004; 90: 121621.
  • 42
    Dales JP,Garcia S,Carpentier S,Andrac L,Ramuz O,Lavaut MN,Allasia C,Bonnier P,Charpin C. Long-term prognostic significance of neoangiogenesis in breast carcinomas: comparison of Tie-2/Tek, CD105, and CD31 immunocytochemical expression. Hum Pathol 2004; 35: 17683.
  • 43
    Dales JP,Garcia S,Meunier-Carpentier S,Andrac-Meyer L,Haddad O,Lavaut MN,Allasia C,Bonnier P,Charpin C. Overexpression of hypoxia-inducible factor HIF-1alpha predicts early relapse in breast cancer: retrospective study in a series of 745 patients. Int J Cancer 2005; 116: 7349.
  • 44
    Meunier-Carpentier S,Dales JP,Djemli A,Garcia S,Bonnier P,Andrac-Meyer L,Lavaut MN,Allasia C,Charpin C. Comparison of the prognosis indication of VEGFR-1 and VEGFR-2 and Tie2 receptor expression in breast carcinoma. Int J Oncol 2005; 26: 97784.
  • 45
    Gradishar WJ,Tjulandin S,Davidson N,Shaw H,Desai N,Bhar P,Hawkins M,O'Shaughnessy J. Phase III trial of nanoparticle albumin-bound paclitaxel compared with polyethylated castor oil-based paclitaxel in women with breast cancer. J Clin Oncol 2005; 23: 7794803.
  • 46
    Kurt M,Harputluoglu H,Dede DS,Gullu IH,Altundag K. Potential molecular targeted therapies in the management of the basal-like subtype of breast cancer. Breast 2007; 16: 1112.
  • 47
    Finn RS,Dering J,Ginther C,Wilson CA,Glaspy P,Tchekmedyian N,Slamon DJ. Dasatinib, an orally active small molecule inhibitor of both the src and abl kinases, selectively inhibits growth of basal-type/“triple-negative” breast cancer cell lines growing in vitro. Breast Cancer Res Treat 2007; 105: 31926.
  • 48
    Quintas-Cardama A,Kantarjian H,Jones D,Nicaise C,O'Brien S,Giles F,Talpaz M,Cortes J. Dasatinib (BMS-354825) is active in Philadelphia chromosome-positive chronic myelogenous leukemia after imatinib and nilotinib (AMN107) therapy failure. Blood 2007; 109: 4979.
  • 49
    Christensen JG,Schreck R,Burrows J,Kuruganti P,Chan E,Le P,Chen J,Wang X,Ruslim L,Blake R,Lipson KE,Ramphal J, et al. A selective small molecule inhibitor of cMet kinase inhibits cMet-dependent phenotypes in vitro and exhibits cytoreductive antitumor activity in vivo. Cancer Res 2003; 63: 734555.
  • 50
    Anfosso F,Bardin N,Vivier E,Sabatier F,Sampol J,Dignat-George F. Outside-in signaling pathway linked to CD146 engagement in human endothelial cells. J Biol Chem 2001; 276: 15649.
  • 51
    Yan X,Lin Y,Yang D,Shen Y,Yuan M,Zhang Z,Li P,Xia H,Li L,Luo D,Liu Q,Mann K, et al. A novel anti-CD146 monoclonal antibody, AA98, inhibits angiogenesis and tumor growth. Blood 2003; 102: 18491.