Peptide Profiling Techniques to Detect Thyroid Carcinoma

描述(由申请人提供)：充分诊断、治疗和监测癌症所需的信息是如此复杂，综合使用一组测量方法可能会提供最佳答案。这一概念体现在SELDI-TOF质谱学(MS)多肽图谱中，这是一种基于血清的癌症检测新技术。尽管SELDI到目前为止只产生了低复杂性的光谱，但当对这些模式进行分组分析时，有可能创造出诊断精度与传统生物标记物一样好或更好的学习算法。我们已经开发了一种在磁性反相微珠上捕获多肽的系统，然后是MALDI-TOF MS，以产生越来越复杂但非常可重复性的模式。这有明显的优势，因为更多的展示多肽提供了更多的机会来选择癌症亚型和分期的独特模式(条形码)，并预测和监测临床结果。还采取了极其谨慎的措施，使标本的收集、处理和储存标准化，以避免引入人工制品。MSKCC针对多种恶性肿瘤的试点项目表明，由此获得的多肽模式似乎包含了可能具有直接临床实用价值的信息。该项目的目标是(I)自动化我们的原型血清多肽图谱平台，并实施机器学习方法，使用得到的多肽模式(条形码)进行样本分类[R21]；以及(Ii)在高通量环境中使用定义良好和仔细观察的甲状腺癌患者组来测试“条形码诊断”模型[R33]。R21目标一是自动化血清样本处理和分析；目标二是自动化所有数据处理，检查模式选择和样本类别预测方法，并整合所有软件平台；目标三是开发条形码多肽的常规MALDI-TOF/TOF串联MS测序。R33目的一是确定甲状腺疾病患者血清模式的重复性；目标二是确定可区分甲状腺癌患者和良性甲状腺结节患者的条形码；目标三是评估血清多肽条形码是否能在一大群甲状腺癌幸存者中识别隐匿性转移。

DESCRIPTION (provided by applicant): The information required for adequate diagnosis, treatment and monitoring of cancers is so complex that a panel of measurements, used in sum, may provide the best answers. The concept is embodied in SELDI-TOF mass spectrometric (MS) peptide profiling, an emerging technique for serum based cancer detection. Even though SELDI has thus far only produced low complexity spectra, the patterns, when analyzed as groups, have the potential to create learning algorithms with diagnostic accuracies as good as or better than conventional biomarkers. We have developed a system to capture peptides on magnetic reversed-phase beads, followed by MALDI-TOF MS, to yield increasingly complex, yet very reproducible patterns. This has clear advantages, as more displayed peptides provide more opportunity to select unique patterns ('barcodes') for cancer subtypes and stages, and to predict and monitor clinical outcome. Extreme care has also been taken to standardize specimen collection, handling and storage to avoid the introduction of artifact. Pilot projects at MSKCC with a variety of malignancies suggest that peptide patterns thus obtained appear to hold information that may have direct clinical utility. The goals of this project are to (i) automate our prototype serum peptide profiling platform and implement machine learning methods that use the resulting peptide patterns ('barcodes') for sample classification [R21]; and (ii) to test the 'barcode diagnostic' model in a high-throughput setting, using well defined and carefully observed groups of thyroid carcinoma patients [R33]. R21 aim one is to automate serum sample processing and analysis; aim two is to automate all data processing, to examine pattern selection and sample class prediction methods, and to integrate all software platforms; aim three is to develop routine MALDI-TOF/TOF tandem MS sequencing of 'barcode' peptides. R33 aim one is to define reproducibility of serum patterns in patients with thyroid disease; aim two is to determine barcodes that can distinguish patients with thyroid cancer from those with benign thyroid nodules; aim three is to assess if serum peptidome barcodes can identify occult metastasis in a large group of thyroid cancer survivors.