Ratech H, Litwin S
Department of Pathology, Fox Chase Cancer Center, Philadelphia, Pennsylvania.
Am J Clin Pathol. 1989 May;91(5):583-6. doi: 10.1093/ajcp/91.5.583.
Twenty-four control cases of reactive follicular hyperplasia (RFH) and 28 cases of B-cell non-Hodgkin's malignant lymphoma (ML) were evaluated for surface immunoglobulin (sIg) light chain expression with flow cytometry. Because there was no significant difference between the distribution of the ML kappa/lambda ratios for which kappa/lambda greater than 1 compared with those ML lambda/kappa ratios for which lambda/kappa greater than 1 using the Kolmogorov-Smirnov statistic, the authors treated kappa/lambda and lambda/kappa data symmetrically. By transforming all kappa/lambda ratios to maximum ratios, that is, Rmax = max (kappa/lambda, lambda/kappa), so that all resulting ratios were at least one, the authors defined a new statistic for evaluating sIg light chain restriction. There was a highly significant difference for equality of distribution of the data using the Mann-Whitney test (P less than 0.0001). By choosing cut-points based on the Rmax statistic, the authors were able to separate the RFH and ML cases with sensitivity = 0.96 +/- 0.07 (95% confidence) and with specificity = 0.91 +/- 0.11 (95% confidence). These point estimates of sensitivity and specificity appear to be promising for the clinical utility of the Rmax statistic, even for cases of B-cell ML with discordantly low expression of sIg light chains. The Rmax statistic will facilitate the comparison of data between different laboratories because it does not require prior assignment of cut-points and can be made using a simple nonparametric test.
采用流式细胞术对24例反应性滤泡增生（RFH）对照病例和28例B细胞非霍奇金恶性淋巴瘤（ML）病例进行表面免疫球蛋白（sIg）轻链表达评估。由于使用Kolmogorov - Smirnov统计量比较κ/λ大于1的ML病例与λ/κ大于1的ML病例时，ML的κ/λ比值分布无显著差异，作者对κ/λ和λ/κ数据进行对称处理。通过将所有κ/λ比值转换为最大比值，即Rmax = max(κ/λ, λ/κ)，使所有结果比值至少为1，作者定义了一种评估sIg轻链限制的新统计量。使用Mann - Whitney检验对数据分布的均等性进行检验，结果显示差异具有高度显著性（P小于0.0001）。通过基于Rmax统计量选择切点，作者能够区分RFH和ML病例，敏感性为0.96±0.07（95%置信度），特异性为0.91±0.11（95%置信度）。这些敏感性和特异性的点估计值对于Rmax统计量的临床应用似乎很有前景，即使对于sIg轻链表达不一致且较低的B细胞ML病例也是如此。Rmax统计量将有助于不同实验室之间的数据比较，因为它不需要事先设定切点，并且可以使用简单的非参数检验进行计算。
