Metrology to address ion suppression in multimodal mass spectrometry imaging with application in oncology

这项研究项目的目的是在测量和理解质谱学成像(MSI)中的电离偏差方面提供一种阶段性的改变。MSI是一项重要的新兴技术，它能够绘制数千个分子的图谱，包括代谢物和药物，在质谱仪中被检测为离子。MSI作为一套模式，可以用来分析几乎任何样本类型中的几乎任何分子，因此有可能彻底改变我们评估生命系统的方式。癌症等疾病涉及人体自然过程的破坏，包括细胞新陈代谢。因此，通过绘制每个图像像素中数千种代谢物的图谱，MSI可以为不同癌症的生长和演变提供强大的洞察力，帮助确定新的治疗靶点。尽管有这样的前景，MSI在检测到的离子信号上仍存在未知的偏差。这些可能会导致误导性的观察，并可能带来代价高昂的影响。电离偏差或基质效应包括一系列现象，这些现象可能导致检测到的离子数量与样品中原始缔合分子数量之间的未知关系。这些偏差在生物样品中存在的浓度范围内也可能是非线性的。因此，如果MSI从业者不能确定，例如，检测到的离子强度增加5倍就反映了原始样品代谢物浓度增加了5倍，显然存在显著的障碍。此外，对于数据集的每个像素中的每个离子，这种现象将以不同的未知程度出现。这相当于每幅图像上有超过1,000,000个离子测量值，每个测量值都有不同的(未知)电离偏差。因此，极大地限制了定量的机会，并提供了对内源性代谢物浓度的错误印象。用于表征电离偏向或量化MSI中内源代谢物浓度的典型方法最多只能研究几个分子。目前还没有现有的方法允许对MSI中的电离偏差进行普遍的研究和校正。此外，还没有开发出标准样本来评估这些现象，因此对不同MSI模式之间的这些偏差缺乏了解。该项目旨在为研究和纠正微电子机械系统中的电离偏差奠定坚实的基础。将通过开发适用于研究电离偏差行为的标准样本来追求严格的经验方法。一组分子将被选为：它们与关键病理学的相关性，例如癌症新陈代谢；物理化学性质；与质谱学成像领域的相关性。这些样品将用来表征包括MALDI和DESI MSI在内的多种质谱学成像方式在检测中的偏差。将建立描述这些电离行为的模型，并开发用于评估和转换这些模型的计算方法。将采用多变量和机器学习的方法来评估有关体系的质谱学和物理化学性质的贡献和联系。

The aim of this research project is to provide a step change in the measurement and understanding of ionisation biases in mass spectrometry imaging (MSI). MSI is an important emerging technology which enables the mapping of thousands of molecules, including metabolites and drugs, detected as ions in the mass spectrometry instrument. MSI, as a suite of modalities, can be employed to analyse almost any molecule in almost any sample type and so has the potential to revolutionise how we evaluate living systems. Diseases such as cancer involve the disruption of the bodies' natural processes including cellular metabolism. MSI, in mapping thousands of metabolites in every image pixel, can therefore provide powerful insights into how different cancers grow and evolve, helping identify new targets for treatment. Despite this promise, MSI suffers from unknown biases in detected ion signal. These can lead to misleading observations with potentially costly implications. Ionisation biases, or matrix effects, encompass a range of phenomena which can lead to unknown relationships between the number of detected ions and the original number of associated molecules in the sample. These biases may also be non-linear across concentration ranges present within biological samples. Therefore, if the MSI practitioner cannot be certain that, for example, a 5 fold increase in detected ion intensity reflects a 5 fold increase in the original sample metabolite concentration, it is clear that a significant hurdle is present. Furthermore, this phenomenon will be present to varying unknow extents for every ion in every pixel of a dataset. This corresponds to significantly upwards of 1,000,000 ion measurements per image, each with different (unknown) ionisation bias. Therefore, drastically limiting opportunities for quantitation and providing erroneous impression of endogenous metabolite concentrations. Typical approaches for characterising ionisation biases or quantifying endogenous metabolite concentration in MSI will only study a few molecules at most. Currently there are no existing methods allowing generalized study and correction of ionisation biases in MSI. Additionally, no standard samples have been developed to allow assessment of these phenomena and so there is a lack of understanding of these biases across between MSI modalities. This project aims to produce a robust foundation for the study and correction of ionisation biases in MSI. A rigorous empirical approach will be pursued through the development of standard samples suitable for studying ionisation bias behaviours. A suite of molecules will be selected for: their relevance to critical pathologies e.g. cancer metabolism; physico-chemical properties; relevance to the mass spectrometry imaging field. These samples will used to characterize the biases in detection across multiple mass spectrometry imaging modalities including MALDI and DESI MSI. Models describing these ionisation behaviours will be produced and computational approaches for evaluation and transformation of these models will be developed. Multivariate and machine learning approaches will be employed to evaluate the contribution and association of mass spectral and physico-chemical properties of the systems in question.