Electronic Supplementary Material (ESI) for Analyst. This journal is The Royal Society of Chemistry Electronic Supplementary Information ATR-FTIR spectroscopy coupled with chemometric analysis discriminates normal, borderline and malignant ovarian tissue: classifying subtypes of human cancer Georgios Theophilou,, Kássio M.G. Lima,, Pierre L. Martin-Hirsch,, Helen F. Stringfellow, Francis L. Martin * Centre for Biophotonics, LEC, Lancaster University, Lancaster LA YQ, UK; Department of Obstetrics and Gynaecology, Central Lancashire Teaching Hospitals NHS Foundation Trust, Preston, UK; Institute of Chemistry, Biological Chemistry and Chemometrics, Federal University of Rio Grande do Norte, Natal 97-97, RN-Brazil *Corresponding author: email: f.martin@lancaster.ac.uk; phone: + () ; fax: + () 7 No. of Pages = 7 No. of Tables = No. of Figures = Abbreviations: LGSC: low grade serous carcinoma HGSC: high grade serous carcinoma EC: Endometrioid carcinoma MC: Mucinous carcinoma MT: Mixed tumour CCC: Clear cell carcinoma CS: Carcinosarcoma : Discriminant function RMI: Risk malignancy index S
Table S. Risk Malignancy index (RMI): Women with ovarian cysts or vague abdominal symptoms undergo screening using the Risk malignancy index. This predicts the risk of an ovarian mass being malignant and dictates further surgical or medical management. Feature RMI RMI Ultrasonic: Bilateral lesions Ascities Multilocular cysts Solid areas Metastases No positive ultrasound features= abnormality= abnormalities= No positive ultrasound features= abnormality= abnormalities= Premenopausal Postmenopausal Ca U/ml U/ml RMI= Ultrasound score Menopausal score Ca in U/ML RMI Risk Women (%) Risk of cancer (%) < Low < - Moderate > High 7 S
Table S. Histopathological classification of ovarian epithelial tumours: Descriptive criteria for classification of ovarian carcinomas according to the World Health organization (WHO),. Carcinosarcoma is not included in the main five categories due to its rarity. Carcinoma subtype Serous Mucinous Endometrioid Clear cell Mixed surface Carcinosarcoma Description Composed of cells ranging in appearance from those resembling fallopian tube epithelium in well-differentiated tumours to anaplastic epithelial cells with severe nuclear atypia in poorly differentiated tumours Low grade High Grade Uniform nuclei -fold variability in nuclear size </ high field powers >/ high field powers mitotic figures mitotic figures Prominent nucleoli Small nucleoli Differentiated architecture Undifferentiated growth with papillary growth Numerous psammoma Few psammoma bodies bodies Resembles intestinal or endocervical epithelium Closely resembles the common variant of endometrioid carcinoma of the uterine corpus Composed of glycogen-containing clear cells and hobnail cells and occasionally other histological types Composed of an admixture of two or more of the five major histological types, and the minor component(s) must comprise alone or together at least % of the tumour Composed of both malignant epithelial and homologous (similar to Mullerian duct system) or heterologous (e.g. cartilage, bone, muscle) stromal elements S
Table S: Internal and external algorithm validation: 7% of the spectra were used to train the algorithm, % to test it internally and % to validate it externally. Normal Borderline Cancer Total Train 9 7 778 Validation Test Table S: Selected wavenumbers for SPA- LDA and GA- LDA. These wavenumbers were used to achieve classification of normal, borderline and malignant ovarian tissue. Classification into normal, borderline and malignant ovaries Chemometric Wavenumbers (cm - ) selected analysis SPA-LDA 9, 99,,8,,,, 77,,,,,,,,,,, 8, 8, 77, 8 GA-LDA 9, 98, 987,, 9, 8, 99,,, 8,, 9,,, 9,,,, 7, 9,, 8, 9,,,, 7, 7, 79 Table S: Selected wavenumbers for SPA- LDA and GA- LDA. These wavenumbers were used to achieve classification of ovarian carcinoma subtypes. Ovarian carcinoma subtype classification Chemometric Wavenumbers (cm - ) selected analysis SPA-LDA 9, 99, 8, 7, 8,,, 9,,, 8, 8,, 9,, 97,,,,, 97, 7, 8 GA-LDA 9, 9, 9, 999,, 8,, 8, 99,,, 9,,, 8, 9,, 8,, 8, 9, 8, 8, 89, 9,,,,,,, 7, 89, 9,, 7,, 8, 7, 8, 7, 7, 77, 78 S
Figure S: Optimization of Principal Component and Wavenumber selection for each of the analytical methods for paired comparison. This example uses the HGSC and LGSC classes. PCA$LDA' Classifica<on%rate%(%)% Absorbance(AU)%. Principal%components% Wavenumber%(cm 7 )% SPA$LDA' Cost%.....9.9.8. Variables Variables% Absorbance%(AU)%.8... 8 - Wavenumber%(cm 7 ) % GA$LDA' Cost%..9.8.7. Absorbance(AU)%.8.... Generations Variables% 8 - Wavenumber%(cm 7 ) % S
Figure S: PCA-LDA derived scores plots obtained from paired Page: x - -. -. -. -.8 - -.. x -... -....... :#HGSC# :#LGSC# :#HGSC# :#MC# :#HGSC# x - - - -. -. x - -..... :#HGSC# :#EC# :#HGSC# :#MT# :#HGSC# :#CS# -. -. -. -. -. S
Figure S: PCA-LDA derived scores plots obtained from paired Page:...... -... -. -. -. -. -. 8 samples...... :#LGSC# :#EC# :#LGSC# :#MT# :#LGSC# :#CS#. -. -. -. -.8 8...... -.... -. -. -. -. :#LGSC# :#MC# :#LGSC# :#EC# :MC# -. 8 -. S7
Figure S: PCA-LDA derived scores plots obtained from paired Page: x - - - - -... -. -. x - 8 - - :#EC# :#MT# :#EC# :#CS# :#MC# - x - - - - -8 -. -. -. -. -. -. -. 8 indice das amostras..... -. -. -. :#EC# :#MC# :#MT# :#MC# :#CS# -. S8
Figure S: PCA-LDA derived scores plots obtained from paired Page: 8 x - :#MT#... :#MT# :#CS# -. - -. - -. -8 -. 8... :#CS#.. -. -. S9
Figure S: SPA-LDA derived scores plots obtained from paired Page:.8.......8...... -. -. -. -. -. -. -. -. :#HGSC# :#LGSC# :#HGSC# :#MC# :#HGSC#. -. -.. -. -. -. -. -. -. -. -. -. :#HGSC# :#EC# :#HGSC# :#MT# :#HGSC# :#CS# -. -.7 S
Figure S: SPA-LDA derived scores plots obtained from paired Page:. -. -. -. -.8 -....... -. :#LGSC# :#EC# :#LGSC# :#MT#...8.....8. 8...... -. :#LGSC# :#MC# :#LGSC# -. 8 - -. - :#LGSC# :#CS# -. -.8 :#EC# -. -. -. :MC# -. -. -.8-8 -. S
Figure S: SPA-LDA derived scores plots obtained from paired Page:. -. -. -. -.. :#EC#.. -. -. -. -. -. -. -. -. -.8 :#EC# :#MT# :#CS# :#MC#...9.8.7 -. :#MC# -.8 -. -. -. -. -.8 -. 8 -. - -. - :#EC# :#MT# :#MC# :#CS# -. -. -. - S
Figure S: Dimensional SPA-LDA derived scores plots obtained from paired Page:... :#MT# -. -.8. -. -. - -. -. :#MT# :#CS# -. -. 8.9.8 :#CS#.7... S
Figure S: GA-LDA derived scores plots obtained from paired Page: -. -. :#HGSC# :#LGSC# -. -. -. :#HGSC# :#EC# -. -. -. -.... -. -.....9.8 :#HGSC# :#MC# :#HGSC# -. -.7 -.8 -.9 -. -. - - - x - -....8. :#HGSC# :#MT# :#HGSC# :#CS#.7... S
Figure S: GA-LDA derived scores plots obtained from paired Page:. x -.8. 7 x - :#LGSC# :#MC#...8....9.8.7.. 8 - -. -7-7. :#LGSC# :#EC# :#LGSC# :#MT# :#LGSC# :#CS# 8. :#LGSC#......99.98.97.9 -.7 -.8 -.9 -. -. :#EC# :MC# -8-8. 8 -. -. -. S
Figure S: GA-LDA derived scores plots obtained from paired Page: -.9 -. :#EC# :#MT# -. -. :#EC# -. -. -. -. -. -............8.8.8.8.79.79.78.78.77 :#EC# :#CS#.77 :#MC# -. -. -. -. :#MC# -. -. -. -. -. -.7 -.7 -.8 :#MT# :#MC# :#CS# S
Figure S: GA-LDA derived scores plots obtained from paired Page: -. :#MT# -8. -. -9 :#MT# :#CS# -. -9. 8.. :#CS#..... S7