Baumann I, Nenninger R, Harms H, Zwierzina H, Wilms K, Feller A C, Ter Meulen V, Müller-Hermelink H K
Institute of Pathology, University of Würzburg, Germany.
Am J Clin Pathol. 1996 Jan;105(1):23-30. doi: 10.1093/ajcp/105.1.23.
This report outlines the morphologic classification of acute myeloid (AML: French-American-British FAB classification: M1) and lymphoid (ALL) leukemia by automatic image analysis and the correlation to immunologic and cytochemical classification. The investigation was carried out on Romanowsky-Giemsa stained bone marrow (n = 15) and blood smears (n = 10) from 25 patients with primary acute leukemia. The cases had been classified as of myeloid or lymphoid origin by three hematologic centers using immunochemistry or cytochemistry, but the specimens were submitted to the authors' laboratory without the diagnosis. The nuclear and cytoplasmic pattern of the blast cells were analyzed by a high resolution image analysis system and the measured and calculated cell features were sorted by means of a classifier program (CART). The image analysis classification was then compared with the immunophenotypical and cytochemical classification. Blood blast cells showed nuclear features that were significantly correlated to a myeloid or lymphoid immunophenotype. In contrast, bone marrow blast cells displayed overlapping and therefore nondiscriminating nuclear features. However, by generating a learning data set using the immunophenotypes the classifier program found specific cytoplasmic features that eventually permitted a differentiation into myeloid or lymphoid subtypes. In summary, the authors suggest that high resolution image analysis of leukemic blast cells detect nuclear and cytoplasmic features that are associated with the immunophenotype and therefore with the lineage determination of the cell. With this new objective and reproducible approach of morphologic cell analysis, it might not only be possible to classify blast cells with minimal cellular differentiation, but furthermore to discover prognostic features because the remarkable difference in classification quality between blood and bone marrow blast cells reported in this study, might be of biologic relevance and requires further investigation.
本报告概述了通过自动图像分析对急性髓系白血病（AML：法美英FAB分类：M1）和淋巴细胞白血病（ALL）进行的形态学分类，以及与免疫和细胞化学分类的相关性。对25例原发性急性白血病患者的罗曼诺夫斯基-吉姆萨染色骨髓涂片（n = 15）和血涂片（n = 10）进行了研究。这三个血液学中心已使用免疫化学或细胞化学将这些病例分类为髓系或淋巴系起源，但标本在未给出诊断结果的情况下被提交至作者的实验室。通过高分辨率图像分析系统分析原始细胞的细胞核和细胞质模式，并借助分类程序（CART）对测量和计算出的细胞特征进行分类。然后将图像分析分类结果与免疫表型和细胞化学分类结果进行比较。血原始细胞显示出与髓系或淋巴系免疫表型显著相关的核特征。相比之下，骨髓原始细胞表现出重叠的核特征，因此无法区分。然而，通过使用免疫表型生成一个学习数据集，分类程序发现了特定的细胞质特征，最终实现了向髓系或淋巴系亚型的区分。总之，作者认为对白血病原始细胞进行高分辨率图像分析可检测到与免疫表型相关的细胞核和细胞质特征，从而与细胞的谱系确定相关。通过这种新的客观且可重复的形态学细胞分析方法，不仅有可能对细胞分化程度最低的原始细胞进行分类，而且还有可能发现预后特征，因为本研究中报道的血原始细胞和骨髓原始细胞在分类质量上的显著差异可能具有生物学相关性，需要进一步研究。
