Biomedical Engineering Department, McGill University, Montreal, Quebec, Canada.
Anal Chem. 2011 Jul 15;83(14):5767-74. doi: 10.1021/ac103239f. Epub 2011 Jun 23.
Taguchi design, a statistics-based design of experiment method, is widely used for optimization of products and complex production processes in many different industries. However, its use for antibody microarray optimization has remained underappreciated. Here, we provide a brief explanation of Taguchi design and present its use for the optimization of antibody sandwich immunoassay microarray with five breast cancer biomarkers: CA15-3, CEA, HER2, MMP9, and uPA. Two successive optimization rounds with each 16 experimental trials were performed. We tested three factors (capture antibody, detection antibody, and analyte) at four different levels (concentrations) in the first round and seven factors (including buffer solution, streptavidin-Cy5 dye conjugate concentration, and incubation times for five assay steps) with two levels each in the second round; five two-factor interactions between selected pairs of factors were also tested. The optimal levels for each factor as measured by net assay signal increase were determined graphically, and the significance of each factor was analyzed statistically. The concentration of capture antibody, streptavidin-Cy5, and buffer composition were identified as the most significant factors for all assays; analyte incubation time and detection antibody concentration were significant only for MMP9 and CA15-3, respectively. Interactions between pairs of factors were identified, but were less influential compared with single factor effects. After Taguchi optimization, the assay sensitivity was improved between 7 and 68 times, depending on the analyte, reaching 640 fg/mL for uPA, and the maximal signal intensity increased between 1.8 and 3 times. These results suggest that Taguchi design is an efficient and useful approach for the rapid optimization of antibody microarrays.
田口设计是一种基于统计学的实验设计方法，广泛应用于许多不同行业的产品和复杂生产过程的优化。然而，它在抗体微阵列优化中的应用一直未得到充分重视。本文简要介绍了田口设计，并展示了其在优化五种乳腺癌生物标志物（CA15-3、CEA、HER2、MMP9 和 uPA）的抗体夹心免疫微阵列中的应用。我们进行了两轮连续优化，每轮实验有 16 次试验。第一轮测试了三个因素（捕获抗体、检测抗体和分析物）在四个不同水平（浓度），第二轮测试了七个因素（包括缓冲溶液、链霉亲和素-Cy5 染料缀合物浓度和五个检测步骤的孵育时间），每个因素有两个水平；还测试了五个选定因素对之间的两个因素相互作用。通过图形确定每个因素的净检测信号增加的最佳水平，并通过统计分析确定每个因素的显著性。捕获抗体、链霉亲和素-Cy5 和缓冲液组成的浓度被确定为所有检测的最重要因素；分析物孵育时间和检测抗体浓度仅对 MMP9 和 CA15-3 分别有重要意义。确定了因素对之间的相互作用，但与单因素效应相比，其影响较小。经过田口优化，根据分析物的不同，检测的灵敏度提高了 7 至 68 倍，达到 uPA 的 640 fg/mL，最大信号强度提高了 1.8 至 3 倍。这些结果表明，田口设计是一种快速优化抗体微阵列的有效且有用的方法。
