O'Looney Nichola, Burgess Stewart T G, Chong Kwan Marisa, Ross Alan J, Robb Janine, Forster Thorsten, Beattie John S, Ghazal Peter, Petrik Juraj, Campbell Colin J
Scottish National Blood Transfusion Service (SNBTS), SNBTS National Reference Laboratory, Royal (Dick) Veterinary College, Edinburgh, U.K.
J Immunoassay Immunochem. 2008;29(2):197-209. doi: 10.1080/15321810801888530.
All donor blood samples must be tested pre-transfusion to determine the blood type of donor erythrocytes, based on the ABO typing system. Current methods of testing are well characterised, but require a number of processing steps prior to analysis. In addition, standard testing protocols require additional assays such as hepatitis C and HIV testing be performed separately. We describe and evaluate a protein microarray platform for ABO blood typing that has the potential to be a simple reliable high throughput method, with the added capability for the integration of other important pre-transfusion tests. Sixty seven donor blood samples were incubated on microarrays printed with multiple spotted replicates of blood type antigen specific antibodies. We utilised a hold-out cross validation approach, combined with Receiver Operator Characteristic (ROC) curves to define thresholds within which a sample could be defined as being of a particular blood type. The threshold values from the ROC curve analysis demonstrated an excellent ability to accurately separate samples based on ABO blood type. The results obtained when the thresholds from the training sets were applied to test sets were also very encouraging, with misclassified samples being present in only 2 of the training sets and a mean classification error of 4.28%. When the mean thresholds were applied to the 67 donor samples, 95.5% were correctly blood typed (64 of 67 samples). We have demonstrated the ability of our protein microarray platform to successfully and accurately type human whole blood samples. We believe that this flexible platform provides a strong basis for an integrated approach for combined blood typing and pathogen testing in human whole blood.
所有供血者血样在输血前必须进行检测,以根据ABO血型系统确定供血者红细胞的血型。目前的检测方法已得到充分表征,但在分析前需要多个处理步骤。此外,标准检测方案要求如丙型肝炎和艾滋病毒检测等额外检测需单独进行。我们描述并评估了一种用于ABO血型分型的蛋白质微阵列平台,该平台有可能成为一种简单可靠的高通量方法,还具备整合其他重要输血前检测的能力。67份供血者血样在印有血型抗原特异性抗体多个斑点复制品的微阵列上进行孵育。我们采用留出法交叉验证方法,结合受试者工作特征(ROC)曲线来确定可将样本定义为特定血型的阈值。ROC曲线分析得出的阈值显示出基于ABO血型准确区分样本的出色能力。将训练集的阈值应用于测试集时获得的结果也非常令人鼓舞,只有2个训练集中存在误分类样本,平均分类误差为4.28%。当将平均阈值应用于67份供血者样本时,95.5%的样本血型分型正确(67份样本中的64份)。我们已经证明了我们的蛋白质微阵列平台成功且准确地对人类全血样本进行血型分型的能力。我们相信,这个灵活的平台为人类全血中血型分型和病原体检测相结合的综合方法提供了坚实基础。