Electronic Supplementary Material
-
In situ validation of VEGFR-2 and αvß3 integrin
as targets for breast lesion characterization
Josef Ehling1,2 §, Matthias Misiewicz1 §, Saskia von Stillfried2, Diana Möckel1, Jessica Bzyl1, Sibylle Pochon3, Wiltrud Lederle1, Ruth Knuechel2, Twan Lammers1,4,5, Moritz Palmowski1, Fabian Kiessling1*
§ Equal contribution
1Department of Experimental Molecular Imaging, Helmholtz Institute for Biomedical Engineering, Medical Faculty, RWTH Aachen University, Aachen, Germany
2Institute of Pathology, Medical Faculty, RWTH Aachen University, Aachen, Germany
3Bracco Suisse SA,Geneva, Switzerland
4Department of Targeted Therapeutics, MIRA Institute for Biomedical Technology and Technical
Medicine, University of Twente, Enschede, The Netherlands
5Department of Pharmaceutics, Utrecht Institute for Pharmaceutical Sciences, Utrecht University,
Utrecht, The Netherlands
* Corresponding author:Prof. Dr. Fabian Kiessling
Department of Experimental Molecular Imaging
Medical Faculty of the RWTH Aachen
Pauwelsstr. 30
52074 Aachen
Germany
Phone: +49-241-8080116
Fax: +49-241-8082006
Email:
Electronic supplementary material:7 Supplementary Figures,
8 Supplementary Tables
Supplementary Tables
Supplementary Table S1: Summary of sensitivity and specificity for differentiating benign (fibroadenoma) from malignant (invasive cancer) breast lesions based on different threshold levels for the filled CD31 area fraction. CI = confidence interval.
Filled CD31 area fraction threshold level [%] / Sensitivity (95% CI) / Specificity (95% CI)3.563 / 97.5 (87 – 100) / 0 (0 – 9)
4.702 / 95 (83 – 99) / 5 (1 – 17)
5.120 / 90 (76 – 97) / 5 (1 – 17)
5.304 / 85 (70 – 94) / 10 (3 – 24)
5.471 / 80 (64 – 91) / 10 (3 – 24)
5.661 / 75 (59 – 87) / 12.5 (4 – 27)
5.965 / 70 (53 – 83) / 20 (9 – 36)
6.664 / 65 (48 – 79) / 30 (17 – 47)
7.190 / 60 (43 – 75) / 40 (25 – 57)
7.383 / 57.5 (41 – 73) / 50 (34 – 66)
7.710 / 55 (38 – 71) / 55 (38 – 71)
7.947 / 50 (34 – 66) / 57.5 (41 – 73)
8.713 / 40 (25 – 57) / 75 (59 – 87)
9.185 / 30 (17 – 47) / 87.5 (73 – 96)
9.848 / 20 (9 – 36) / 90 (76 – 97)
10.79 / 10 (3 – 24) / 97.5 (87 – 100)
Supplementary Table S2: Summary of sensitivity and specificity for differentiating normal breast tissue from benign breast lesions (fibroadenoma) based on different threshold levels for the filled CD31 area fraction.
Filled CD31 area fraction threshold level [%] / Sensitivity (95% CI) / Specificity (95% CI)2.388 / 100 (91 – 100) / 2.5 (0 – 13)
2.463 / 100 (91 – 100) / 10 (3 – 24)
2.932 / 100 (91 – 100) / 20 (9 – 36)
3.470 / 100 (91 – 100) / 30 (17 – 47)
3.693 / 100 (91 – 100) / 40 (25 – 57)
4.323 / 97.5 (87 – 100) / 50 (34 – 66)
4.976 / 95 (83 – 99) / 60 (43 – 75)
5.446 / 90 (76 – 97) / 65 (48 – 79)
6.330 / 72.5 (56 – 85) / 72.5 (56 – 85)
6.613 / 70 (53 – 83) / 75 (59 – 87)
7.153 / 60 (43 – 75) / 77.5 (62 – 89)
7.383 / 50 (34 – 66) / 77.5 (62 – 89)
8.120 / 40 (25 – 57) / 90 (76 – 97)
8.242 / 30 (17 – 47) / 90 (76 – 97)
9.819 / 20 (9 – 36) / 90 (76 – 97)
9.849 / 10 (3 – 24) / 92.5 (80 – 98)
Supplementary Table S3: Summary of sensitivity and specificity for differentiating benign (fibroadenoma) from malignant (invasive cancer) breast lesions based on different VEGFR-2 area fraction threshold levels.
VEGFR-2 area fraction threshold level [%] / Sensitivity (95% CI) / Specificity (95% CI)0.02185 / 100 (91 - 100) / 10 (3 – 24)
0.03615 / 95 (83 – 99) / 20 (9 – 36)
0.05571 / 90 (76 – 97) / 40 (25 – 57)
0.06000 / 85 (70 – 94) / 45 (29 – 61)
0.0640 / 80 (64 – 91) / 47.5 (32 – 64)
0.06815 / 75 (59 – 87) / 55 (38 – 71)
0.07746 / 70 (53 – 83) / 60 (43 – 75)
0.08999 / 67.50 (51 – 81) / 70 (53 – 83)
0.09213 / 65 (48 – 79) / 72.5 (56 – 85)
0.09613 / 60 (43 – 75) / 72.5 (56 – 85)
0.1055 / 55 (38 – 71) / 75 (59 – 87)
0.1195 / 50 (34 – 66) / 75 (59 – 87)
0.1348 / 40 (25 – 57) / 82.5 (67 – 93)
0.1534 / 30 (17 – 47) / 90 (76 – 97)
0.1909 / 20 (9 – 35) / 92.5 (80 – 98)
0.2301 / 10 (3 – 24) / 100 (91 – 100)
Supplementary Table S4: Summary of sensitivity and specificity for differentiating normal breast tissue from benign breast lesions (fibroadenoma) based on different VEGFR-2 area fraction threshold levels.
VEGFR-2 area fraction threshold level [%] / Sensitivity (95% CI) / Specificity (95% CI)0.0145 / 100 (91 - 100) / 5 (1 – 17)
0.01715 / 95 (83 – 99) / 7.5 (2 – 20)
0.02065 / 90 (76 – 97) / 15 (6 – 30)
0.02545 / 87.5 (73 – 96) / 25 (13 – 41)
0.0316 / 85 (70 – 94) / 45 (29 – 62)
0.03625 / 80 (64 – 91) / 55 (38 – 71)
0.0424 / 75 (59 – 87) / 60 (43 – 75)
0.04735 / 70 (53 – 83) / 65 (48 – 79)
0.04805 / 65 (48 – 79) / 65 (48 – 79)
0.05767 / 60 (43 – 75) / 70 (53 – 83)
0.06495 / 50 (34 – 66) / 75 (59 – 87)
0.06495 / 50 (34 – 66) / 75 (59 – 87)
0.0795 / 40 (25 – 57) / 85 (70 – 94)
0.09015 / 30 (17 – 47) / 87.5 (73 – 96)
0.1313 / 20 (9 – 35) / 100 (91 – 100)
0.1575 / 10 (3 – 24) / 100 (91 – 100)
Supplementary Table S5: Summary of sensitivity and specificity for differentiating benign (fibroadenoma) from malignant (invasive cancer) breast lesions based on different threshold levels for the endothelial αvß3area fraction.
Endothelial αvß3 area fraction threshold level [%] / Sensitivity (95% CI) / Specificity (95% CI)0.04135 / 100 (91 - 100) / 20 (9 – 36)
0.04840 / 95 (83 – 99) / 22.5 (11 – 38)
0.05342 / 90 (76 – 97) / 30 (17 – 47)
0.05640 / 85 (70 – 94) / 32.5 (17 – 49)
0.05966 / 80 (64 – 91) / 45 (29 – 62)
0.06120 / 75 (59 – 87) / 47.5 (32 – 64)
0.06214 / 70 (53 – 83) / 47.5 (32 – 64)
0.06487 / 67.50 (51 – 81) / 62.5 (46 – 77)
0.06548 / 65 (48 – 79) / 62.5 (46 – 77)
0.06888 / 60 (43 – 75) / 67.5 (51 – 81)
0.0704 / 55 (38 – 71) / 70 (53 – 83)
0.07083 / 50 (34 – 66) / 70 (53 – 83)
0.07831 / 40 (25 – 57) / 82.5 (67 – 93)
0.08891 / 30 (17 – 47) / 82.5 (67 – 93)
0.1005 / 20 (9 – 36) / 95 (83 – 99)
0.1257 / 10 (3 – 24) / 100 (91 – 100)
Supplementary Table S6: Summary of sensitivity and specificity for differentiating normal breast tissue from benign breast lesions (fibroadenoma) based on different threshold levels for the endothelial αvß3area fraction.
Endothelial αvß3 area fraction threshold level [%] / Sensitivity (95% CI) / Specificity (95% CI)0.01065 / 100 (91 - 100) / 2.5 (1 – 13)
0.01670 / 97.5 (87 – 100) / 10 (3 – 24)
0.02831 / 95 (83 – 99) / 22.5 (11 – 38)
0.03031 / 90 (76 – 97) / 27.5 (15 – 44)
0.03653 / 85 (70 – 94) / 40 (25 – 57)
0.04078 / 80 (64 – 91) / 50 (34 – 66)
0.04929 / 77.5 (62 – 89) / 67.5 (51 – 81)
0.05362 / 70 (53 – 83) / 70 (53 – 83)
0.05823 / 65 (48 – 79) / 82.5 (67 – 93)
0.05923 / 60 (43 – 75) / 87.5 (73 – 96)
0.05990 / 55 (38 – 71) / 90 (76 – 97)
0.06348 / 50 (34 – 66) / 95 (83 – 99)
0.06463 / 40 (25 – 57) / 95 (83 – 99)
0.07046 / 30 (17 – 47) / 97.5 (87 – 100)
0.07613 / 20 (9 – 36) / 97.5 (87 – 100)
0.09425 / 10 (3 – 24) / 100 (91 – 100)
Supplementary Table S7: Summary of sensitivity and specificity for differentiating benign (fibroadenoma) from malignant (invasive cancer) breast lesions based on different threshold levels for the global αvß3area fraction.
Total αvß3 area fraction threshold level [%] / Sensitivity (95% CI) / Specificity (95% CI)2.871 / 100 (91 - 100) / 20 (9 – 36)
2.936 / 95 (83 – 99) / 20 (9 – 36)
3.543 / 92.5 (80 – 98) / 50 (34 – 66)
3.611 / 90 (76 – 97) / 50 (34 – 66)
3.836 / 85 (70 – 94) / 60 (43 – 75)
4.062 / 82.5 (67 – 93) / 70 (53 – 83)
4.090 / 80 (64 – 91) / 70 (53 – 83)
4.190 / 75 (59 – 87) / 72.5 (56 – 85)
4.230 / 70 (53 – 83) / 72.5 (56 – 85)
4.269 / 65 (48 – 79) / 72.5 (56 – 85)
4.316 / 60 (43 – 75) / 72.5 (56 – 85)
4.456 / 55 (38 – 71) / 80 (64 – 91)
4.486 / 50 (34 – 66) / 82.5 (67 – 93)
4.891 / 40 (25 – 57) / 85 (70 – 94)
5.418 / 30 (17 – 47) / 90 (76 – 98)
5.788 / 20 (9 – 36) / 92.5 (80 – 98)
Supplementary Table S8: Summary of sensitivity and specificity for differentiating normal breast tissue from benign breast lesions (fibroadenoma) based on different threshold levels for the global αvß3area fraction.
Total αvß3 area fraction threshold level [%] / Sensitivity (95% CI) / Specificity (95% CI)1.506 / 97.5 (87 - 100) / 2.5 (0 – 13)
1.860 / 95 (83 – 99) / 10 (3 – 24)
2.205 / 90 (76 – 97) / 20 (9 – 36)
2.437 / 85 (70 – 94) / 30 (17 – 47)
2.511 / 85 (70 – 94) / 40 (24 – 57)
2.641 / 85 (70 – 94) / 50 (34 – 66)
2.824 / 80 (64 – 91) / 65 (48 – 79)
2.989 / 75 (59 – 87) / 75 (59 – 87)
3.211 / 65 (48 – 79) / 75 (59 – 87)
3.270 / 60 (43 – 75) / 80 (64 – 91)
3.444 / 55 (37 – 71) / 85 (70 – 94)
3.604 / 50 (34 – 66) / 90 (76 – 97)
3.853 / 40 (25 – 57) / 95 (83 – 99)
4.113 / 30 (17 – 47) / 100 (91 – 100)
4.447 / 20 (9 – 36) / 100 (91 – 100)
5.310 / 10 (3 – 24) / 100 (91 – 100)
Supplementary Figure Legends
Supplementary Figure S1.Work flow for histologically quantifying the VEGFR-2 and αvß3 integrin expression as well as the vascular area fraction in human tissue samples. (a) Quantification of the endothelial VEGFR-2area fractionupon performing immunofluorescent co-stainings for VEGFR-2 (red), CD31 (green) and cell nuclei (Hoechst; blue) (left panel) and threshold-based segmentation of the total VEGFR-2 fluorescence signal (middle panel). Non-endothelial signal segmentations, which did not match with CD31 co-stainings, were manually excluded to specifically quantify the endothelial VEGFR-2 area fraction (right panel). Arrows indicate endothelial VEGFR-2, * indicates VEGFR-2 aberrantly expressed by tumor cells or tumor-associated stromal cells. (b) Quantification of the total and endothelial αvß3 integrinarea fraction upon performing immunofluorescent co-stainings for αvß3 integrin(red), CD31 (green) and cell nuclei (Hoechst; blue) (left panel) and threshold-based segmentation of the total αvß3 integrinfluorescence signal (middle panel). Subsequently, the total αvß3 integrin area fraction (upper right panel) was quantified without any post-processing and the endothelial αvß3 integrin area (lower right panel) was quantified as described in (a). Arrows indicate endothelial αvß3 integrin, # indicates αvß3 integrin expressed by tumor cells. (c) Quantification of the filled CD31 area fraction upon manually highlighting all CD31-positive vascular structures (cf. left and middle panel). Subsequently, highlighted vessel borders were computationally filled using a custom macro implemented for open-access ImageJ analysis software (right panel).
Supplementary Figure S2.Vascular area fraction as well as VEGFR-2 and αvß3 integrin expression in radial scars and DCIS lesions of the human breast.(a) Representative photomicrographs of radial scars (left column) and ductal carcinoma in situ (DCIS) lesions (right column) co-stained with Hoechst cell nuclei marker (blue, first row), against endothelial cell marker CD31 (green, second row) and against VEGFR-2 (red, third row). Merged images (4th column) demonstrate a high endothelial VEGFR-2 expression in pre-invasive DCIS lesions and a low expression in radial scars. Quantification of the filled CD31 (b) and endothelial VEGFR-2 (c) area fraction in both cohorts. (d)Representative photomicrographs of radial scars (left column) and ducal carcinoma in situ (DCIS) lesions (right column) co-stained with Hoechst cell nuclei marker (blue, first row), against endothelial cell marker CD31 (green, second row) and against αvß3 integrin (red, third row). Merged images (4th column) demonstrate a high total αvß3 expression in pre-invasive DCIS and a relatively low expression in radial scars.Quantification of the total (e) and the endothelial αvß3 integrin (f) area fraction in both cohorts. Data are shown as scattered dot plots including median with interquartile range; n=6 patients for radial scars and n=8 patients for DCIS lesions (Kruskal-Wallis test including Dunn's multiple comparison test).
Supplementary Figure S3. VEGFR-2 and αvß3 integrin expression in ductal vs. lobular breast cancer. (a-d) Quantification of the filled CD31 (a), endothelial VEGFR-2 (b), total αvß3 integrin (c) and endothelial αvß3 integrin (d) area fraction in ductal (non-filled dots) and lobular carcinoma (filled dots). Data are shown as scattered dot plots including median with interquartile range; n=20 patients per group (Mann-Whitney test).
Supplementary Figure S4. Vascular area fraction as well as VEGFR-2 and αvß3 integrinexpression in carcinomas grouped by tumor grading. Quantification of the filled CD31 (upper left panel), endothelial VEGFR-2 (upper right panel), endothelial αvß3 integrin (lower left panel) and total αvß3 integrin (lower right panel) area fraction grouped by tumor grading (G1: low grade; G2: intermediate grade; G3: high grade). Data are shown as scattered dot plots including median with interquartile range; n=2 patients for G1, n=25 patients for G2 and n=13 patients for G3 (Kruskal-Wallis test including Dunn's multiple comparison test).
Supplementary Figure S5. Vascular area fraction as well as VEGFR-2 and αvß3 integrin expression in carcinomas grouped by tumor size. Quantification of the filled CD31 (upper left panel), endothelial VEGFR-2 (upper right panel), endothelial αvß3 integrin (lower left panel) and total αvß3 integrin (lower right panel) area fraction grouped by tumor size (tumor size ≤ 2cm vs tumor size > 2cm). Data are shown as scattered dot plots including median with interquartile range; n=24 patients for ≤ 2cm and n=16 patients for > 2cm (Mann-Whitney test).
Supplementary Figure S6. Correlation of the filled CD31 area fraction as a measure of tissue vascularization and VEGFR-2 area fraction representing angiogenic activity.Significant correlations were found between filled CD31 and VEGFR-2 area fractions for the carcinoma (right panel) and fibroadenoma cohort (middle panel), but not for patients with normal breast tissue (left panel). Correlation analyses were performed by calculating Pearson’s correlation coefficient (r).
Supplementary Figure S7.ROC analyses for endothelialαvß3 integrin expression.ROC analyses based on the endothelial αvß3 integrin signalfor differentiating normal breast vs. fibroadenoma (left panels), normal breast vs. carcinoma (middle panels) and fibroadenoma vs. carcinoma (right panels; n=40 patients per group).
Supplementary Figure S1
Supplementary Figure S2
Supplementary Figure S3
Supplementary Figure S4
Supplementary Figure S5
Supplementary Figure S6
Supplementary Figure S7
1