Moeder Christopher B, Giltnane Jennifer M, Harigopal Malini, Molinaro Annette, Robinson Andrew, Gelmon Karen, Huntsman David, Camp Robert L, Rimm David L
Department of Pathology, Yale University School of Medicine, New Haven, CT 06520-8023, USA.
J Clin Oncol. 2007 Dec 1;25(34):5418-25. doi: 10.1200/JCO.2007.12.8033.
The variability in scoring of immunohistochemistry, whether a result of true heterogeneity or artifacts in preparation, has led to decreased reliability in companion diagnostics and the recommendation for new standards (eg, the American Society of Clinical Oncology/College of American Pathologists [ASCO-CAP] guidelines). The basis of this problem is the amount of tissue required to be representative of an entire tumor. Because protein expression on tissue microarrays (TMAs) can be rigorously measured and one 0.6-mm spot is equivalent to two to three high-power fields, we used TMAs to assess levels of heterogeneity and to determine optimal representation as a function of outcome.
We analyzed estrogen receptor (ER), progesterone receptor, and human epidermal growth factor receptor 2 (HER-2) expression in two cohorts (n = 676 and n = 152) on a series of four to five separate TMA cores and assessed heterogeneity by linear regression analysis. Minimum, average, and maximum scores were generated for each set, which were then assessed for prognostic and predictive value.
Each marker shows some heterogeneity, but average r values between 0.7 and 0.8 are seen between TMA spots. Analysis for prognostic value shows that the highest maximum score (of five spots) is the most prognostic for ER, whereas a high HER-2 minimum score is most prognostic for poor outcome and most predictive of response to trastuzumab.
These results suggest that the representivity required for each biomarker may be a function of its role in tumorigenesis. Furthermore, these results provide scientific basis for the ASCO-CAP guidelines for assessment of HER-2 expression but perhaps suggest that the 30% figure is still too conservative.
免疫组化评分的变异性，无论源于真正的异质性还是制备过程中的假象，都已导致伴随诊断的可靠性降低，并催生了对新标准的需求（例如美国临床肿瘤学会/美国病理学家协会[ASCO-CAP]指南）。这一问题的根源在于需要足量的组织才能代表整个肿瘤。由于组织微阵列（TMA）上的蛋白质表达能够得到精确测量，且一个0.6毫米的斑点相当于两到三个高倍视野，我们利用TMA来评估异质性水平，并根据结果确定最佳代表性。
我们在一系列四到五个单独的TMA核心样本上，分析了两个队列（n = 676和n = 152）中的雌激素受体（ER）、孕激素受体和人表皮生长因子受体2（HER-2）表达情况，并通过线性回归分析评估异质性。为每组生成最低、平均和最高得分，然后评估其预后和预测价值。
每个标志物均显示出一定程度的异质性，但TMA斑点之间的平均r值在0.7至0.8之间。预后价值分析表明，（五个斑点中的）最高最高得分对ER的预后作用最强，而HER-2的最低高分对不良预后的指示作用最强，且对曲妥珠单抗反应的预测性也最强。
这些结果表明，每种生物标志物所需的代表性可能与其在肿瘤发生中的作用有关。此外，这些结果为ASCO-CAP评估HER-2表达的指南提供了科学依据，但或许也表明30%这一数字仍然过于保守。
