CLINICAL AND LABORATORY INVESTIGATIONS DOI 10.1111/j.1365-2133.2007.08185.x Differentiation of tumour-stage mycosis fungoides, psoriasis vulgaris and normal controls in a pilot study using serum proteomic analysis E.W. Cowen, C-W. Liu,* S.M. Steinberg, S. Kang,à E.C. Vonderheid,§ H.S. Kwak,à S. Booher, E.F. Petricoin,– L.A. Liotta,– G. Whiteley* and S.T. Hwang Dermatology Branch, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892, U.S.A. *SAIC-Frederick, Inc., Clinical Proteomics Reference Laboratory, Gaithersburg, MD, U.S.A.  Biostatistics and Data Management Section, Center for Cancer Research, National Cancer Institute, Rockville, MD, U.S.A. àDepartment of Dermatology, University of Michigan Medical School, Ann Arbor, MI, U.S.A. §Departments of Dermatology and Oncology, Johns Hopkins Medical Institutes, Baltimore, MD, U.S.A. –Center for Applied Proteomics and Molecular Medicine, Department of Molecular and Microbiology, George Mason University, Manassas, VA, U.S.A. Correspondence Edward W. Cowen. E-mail: cowene@mail.nih.gov Accepted for publication 24 May 2007 Key words mycosis fungoides, proteomics, psoriasis Conflicts of interest None declared. Summary Background Serum proteomic analysis is an analytical technique utilizing high- throughput mass spectrometry (MS) in order to assay thousands of serum proteins simultaneously. The resultant ‘proteomic signature’ has been used to differentiate benign and malignant diseases, enable disease prognosis, and moni- tor response to therapy. Objectives This pilot study was designed to determine if serum protein patterns could be used to distinguish patients with tumour-stage mycosis fungoides (MF) from patients with a benign inflammatory skin condition (psoriasis) and ⁄ or sub- jects with healthy skin. Methods Serum was analysed from 45 patients with tumour-stage MF, 56 patients with psoriasis, and 47 controls using two MS platforms of differing resolution. An artificial intelligence-based classification model was constructed to predict the presence of the disease state based on the serum proteomic signature. Results Based on data from an independent testing set (14–16 subjects in each group), MF was distinguished from psoriasis with 78Æ6% (or 78Æ6%) sensitivity and 86Æ7% (or 93Æ8%) specificity, while sera from patients with psoriasis were distinguished from those of nonaffected controls with 86Æ7% (or 93Æ8%) sensitiv- ity and 75Æ0% (or 76Æ9%) specificity (depending on the MS platform used). MF was distinguished from unaffected controls with 61Æ5% (or 71Æ4%) sensitivity and 91Æ7% (or 92Æ9%) specificity. In addition, a secondary survival analysis using 11 MS peaks identified significant survival differences between two MF groups (all P-values <0Æ05). Conclusions Serum proteomics should be further investigated for its potential to identify patients with neoplastic skin disease and its ability to determine disease prognosis. Serum proteomics, the study of a subset of circulating pro- teins, is a novel method to study complex human disease such as cancer. Over the last several years, several reports have described the use of proteomic pattern analysis to determine new biomarkers of disease, 1,2 enable disease prognosis 3,4 and monitor response to therapy. 5 Human skin is continuously perfused by blood, and its cellular components are bathed in extracellular fluid. It is therefore likely that proteins associated with neoplastic or inflammatory processes in the skin are released into the extracellular space, entering the lymphatic vessels, and, ultimately, the blood vascular system via the tho- racic duct. Endothelial cells in skin also produce a variety of cytokines, chemokines and adhesion molecules that are released directly into the circulation during inflammatory skin reactions. Both of these processes may be expected to lead to quantitative as well as qualitative changes in the serum protein Journal Compilation Ó 2007 British Association of Dermatologists • British Journal of Dermatology 2007 157, pp946–953 946 No claim to original US government works