(131a) Ovarian Cancer Detection: Understanding Proteomic Data through Chemometric Approaches

In 2007, about 15,280 women in the USA will die because of ovarian cancer and there will be about 22,430 new cases as estimated by the American Cancer Society. Ovarian cancer is insidious, producing very few symptoms until it has spread beyond the ovaries, which precludes complete surgical removal, and accounts for an overall cure rate of only 35%. However, the cure rate is 90% if ovarian cancer can be diagnosed at early stage (National Cancer Institute, 2007). Therefore, ovarian cancer early detection remains the most promising approach to dramatically reduce the mortality associated with this disease. However, finding biomarkers for early cancer detection is very challenging because of modest differences between normal tissue and tumor and large variability in the level of the biomarkers in the control or diseased tissues (Wagner et al., 2004).

Recent advances in mass spectrometry (MS), such as surface-enhanced laser desorption ionization (SELDI), hold great promise for early ovarian cancer detection through proteomic profiling of patient serum. Because thousands of proteins and peptides can be characterized and quantified at the same time, large amount of valuable data are obtained for identifying characteristic and effective biomarkers for ovarian cancer detection. Several advanced data mining algorithms have been reported to be promising for diagnosis of early-stage ovarian cancer (Petricoin et al., 2002; Wulfkuhle et al., 2003; Tirumalai et al., 2003; Zhu et al., 2003; Yanagisawa et al., 2003; Pan et al., 2005). However, considerable controversy has been generated, and there remain some critical issues such as reproducibility and robustness of these methods, which make the proteomic profiling approach has yet to be established (Diamandis 2004; Baggerly et al., 2005; Ransohoff, 2005).

Most published data mining methods are pattern recognition methods. It has been shown that some reported classification results cannot be reproduced (Baggerly et al. 2005). In this work, we consider the ovarian cancer detection problem from a chemical engineering perspective, i.e., treat the cancer detection problem as a fault detection problem in human body, and apply principles and techniques developed for fault detection in chemical engineering to the analysis of a publicly available ovarian cancer dataset (National Cancer Institute Dataset 08-07-02). We apply chemometric approaches, such as Principal Component Analysis (PCA) and Discriminant Partial Least Squares (DPLS), to the dataset. Unique characteristics of the ovarian proteomic data observed through the study are discussed. We also compare the results obtained using chemometric methods with those obtained from pattern recognition methods such as Fisher Discriminant Analysis (FDA).

Key words:

Proteomic data, ovarian cancer detection, chemometric methods

Reference:

1. Baggerly K.A., Morris J.S., Edmonson S.R., Coombes K.R. (2005), Signal in noise: evaluating reported reproducibility of serum proteomic tests for ovarian cancer, J Natl Cancer Inst, Vol. 97 (4), 307 ? 309

2. Liotta L.A., Lowenthal M., Mehta A., Conrads T.P., Veenstra, T.D., Fishman D.A., Petricoin E.F. (2005), Importance of communication between producers and consumers of publicly available experimental data, J Natl Cancer Inst, Vol. 97 (4), 310 ? 314.

3. Ransohoff D.F. (2005), Lessons from controversy: ovarian cancer screening and serum proteomics. J Natl Cancer Inst, Vol. 97 (4), 315 ? 319.

4. Wolkenhauer O., Ghosh B.K., and Cho K.H. (2004), Control and coordination in biochemical networks. IEEE Control Systems Magazine, 24(4):30-34.

5. Diamandis E.P. (2004), Analysis of serum proteomic patterns for early cancer diagnosis: drawing attention to potential problems. J Natl Cancer Inst, 96(5):353 ? 356.

6. National cancer institute (2007), Women's Health Report, Fiscal Years 2005-2006

7. Pan S., Zhang H., Rush J., Eng J., Zhang N, Patterson D., Comb M.J., and Aebersold R. (2005), High throughput proteome screening for biomarker detection. Molecular & Cellular Proteomics, 4(2):182 ? 190.

8. Petricoin E.F., Ardekani A.M., Hitt B.A., Levine P.J., Fusaro V.A., Steinberg S.M., Mills G.B., Simone C., Fishman D.A., Kohn E.C., Liotta A.A. (2002), Use of proteomic patterns in serum to identify ovarian cancer, Lancet, Vol. 359, pp572 - 7.

9. Tirumalai R.S., Chan K.C., Prieto D.A., Issaq H.J., Conrads T.P., and Veenstra T.D. (2003), Charaterization of the low molecular weight human serum proteome. Molecular & Cellular Proteomics, 2:1096 ? 1103.

10. Wu B., Abbott T., Fishman D., McMurray W., Mor G., Stone K., Ward D., Williams K., Zhao H. (2003), Comparison of statistical methods for classification of ovarian cancer using mass spectrometry data, Bioinformatics, Vol. 19 (13), 1636 ? 43.

11. Wulfkuhle J.D., Liotta L.A., and Petricoin E.F. (2003) Proteomic applications for the early detection of cancer. Nature Reviews Cancer, 3(4):267 ? 275.

12. Yanagisawa K., Shyr Y., Xu B.J., Massion P.P., Larsen P.H., White B.C., Roberts J.R., Edgerton M., Gonzalez A., Nadaf S., Moore J.H, Caprioli R.M., and Carbone D.P. (2003), Proteomic patterns of tumour subsets in non-small-cell lung cancer. Lancet, 9382(9):433 ? 439.

13. Zhang Z., Bast R.C., Yu Y., Li J., Sokoll L.J., Rai A.J., Rosenzweig J.M., Cameron B., Wang Y.Y., Meng X., Berchuck A., van Haaften-Day C., Hacker N.F., de Burijn H.W.A., van der Zee A.G.J., Jocobs I.J., Fung E.T., Chan D.W. (2004), Three biomarkers identified from serum proteomic analysis for the detection of early stage ovarian cancer, Cancer Research, Vol. 64, 5882 ? 5890.

14. Zhu W., Wang X., Rao M., Glimm J., Kovach J.S. (2003), Detection of cancer-specific markers amid massive mass spectral data, Proceedings of National Academy of Science, Vol. 100 (25), 14666 ? 71