Abstract 1705: Diagnosis of chronic lymphoid leukemia on unstained blood smears using Raman microspectroscopy and supervised classification

Abstract Chronic Lymphoid Leukemia (CLL) is a blood cancerous disease characterized by the proliferation of lymphocytes (lymphocytosis). This is the most common leukemia, preferentially affecting people aged over 50 years old. It is incurable and in most cases shows no clinical signs. Thus, it is often discovered by chance during a blood test. If necessary, morphological and immunological studies are led by analyzing blood smears colored with May-Grünwald Giemsa, by making a complete blood count and by computing a Matutes scores. These studies are necessary because it is impossible to distinguish a healthy cell from a cancerous one only using a conventional microscope even if the cells are stained. In order to establish the diagnosis of the disease reliably and automatically, a new approach based on the Raman microspectroscopy is introduced in this study. Based on the interaction between matter and light, this technology acquires a real molecular and biochemical signature of the analyzed sample in a label-free manner (no staining or marking of the sample). Recently, Raman microspectroscopy has shown a promising potential to highlight the pathophysiological states in both human and animal since differences between different pathophysiological states occur at the molecular level. From unstained blood smears, obtained by spreading a drop of blood of CLL patients and healthy persons on standard glass slides, Raman spectra are acquired on the nucleus of white blood cells. Using a supervised classification algorithm such as the “Support Vector Machine” (SVM), a spectral model is established to distinguish spectra obtained from healthy or pathological cells. This model is computed using a population of spectra called “training set” and consists of 513 spectra from 6 CLL patients and 347 spectra from 4 healthy persons. Then, a classification of the “validation set” composed of 2820 spectra from 33 CLL patients and 1074 spectra from13 healthy individuals was achieved. A sensitivity of 84.84% and a specificity of 84.61% were obtained on the patient. Finally, a “blind set” composed of 721 spectra from 8 cases is classified. The sensitivity and specificity are respectively equal to 80% and 100%. The results obtained on the differentiation between healthy and pathological cells show the real potential of Raman spectroscopy coupled with supervised classification for the diagnosis of CLL. Future studies will concentrate on the prognosis, and if possible the prediction of CLL. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 1705. doi:1538-7445.AM2012-1705