Emrys A. Jones

Imaging mass spectrometry statistical analysis

Imaging mass spectrometry is increasingly used to identify new candidate biomarkers. This clinical application of imaging mass spectrometry is highly multidisciplinary: expertise in mass spectrometry is necessary to acquire high quality data, histology is required to accurately label the origin of each pixels mass spectrum, disease biology is necessary to understand the potential meaning of the imaging mass spectrometry results, and statistics to assess the confidence of any findings. Imaging mass spectrometry data analysis is further complicated because of the unique nature of the data (within the mass spectrometry field); several of the assumptions implicit in the analysis of LC-MS/profiling datasets are not applicable to imaging. The very large size of imaging datasets and the reporting of many data analysis routines, combined with inadequate training and accessible reviews, have exacerbated this problem. In this paper we provide an accessible review of the nature of imaging data and the different strategies by which the data may be analyzed. Particular attention is paid to the assumptions of the data analysis routines to ensure that the reader is apprised of their correct usage in imaging mass spectrometry research.

Multiple Statistical Analysis Techniques Corroborate Intratumor Heterogeneity in Imaging Mass Spectrometry Datasets of Myxofibrosarcoma

MALDI mass spectrometry can generate profiles that contain hundreds of biomolecular ions directly from tissue. Spatially-correlated analysis, MALDI imaging MS, can simultaneously reveal how each of these biomolecular ions varies in clinical tissue samples. The use of statistical data analysis tools to identify regions containing correlated mass spectrometry profiles is referred to as imaging MS-based molecular histology because of its ability to annotate tissues solely on the basis of the imaging MS data. Several reports have indicated that imaging MS-based molecular histology may be able to complement established histological and histochemical techniques by distinguishing between pathologies with overlapping/identical morphologies and revealing biomolecular intratumor heterogeneity. A data analysis pipeline that identifies regions of imaging MS datasets with correlated mass spectrometry profiles could lead to the development of novel methods for improved diagnosis (differentiating subgroups within distinct histological groups) and annotating the spatio-chemical makeup of tumors. Here it is demonstrated that highlighting the regions within imaging MS datasets whose mass spectrometry profiles were found to be correlated by five independent multivariate methods provides a consistently accurate summary of the spatio-chemical heterogeneity. The corroboration provided by using multiple multivariate methods, efficiently applied in an automated routine, provides assurance that the identified regions are indeed characterized by distinct mass spectrometry profiles, a crucial requirement for its development as a complementary histological tool. When simultaneously applied to imaging MS datasets from multiple patient samples of intermediate-grade myxofibrosarcoma, a heterogeneous soft tissue sarcoma, nodules with mass spectrometry profiles found to be distinct by five different multivariate methods were detected within morphologically identical regions of all patient tissue samples. To aid the further development of imaging MS based molecular histology as a complementary histological tool the Matlab code of the agreement analysis, instructions and a reduced dataset are included as supporting information.

Chemo-informatic strategy for imaging mass spectrometry-based hyperspectral profiling of lipid signatures in colorectal cancer

Significance Mass spectrometry imaging (MSI) technology represents a highly promising approach in cancer research. Here, we outline current roadblocks in translational MSI and introduce a comprehensive workflow designed to address current methodological limitations. An integrated bioinformatics platform is presented that allows intuitive histology-directed interrogation of MSI datasets. We show that this strategy permits the analysis of multivariate molecular signatures with direct correlation to morphological regions of interest, which can offer new insights into how different tumor microenvironmental populations interact with one another and generate novel region-of-interest specific biomarkers and therapeutic targets. Mass spectrometry imaging (MSI) provides the opportunity to investigate tumor biology from an entirely novel biochemical perspective and could lead to the identification of a new pool of cancer biomarkers. Effective clinical translation of histology-driven MSI in systems oncology requires precise colocalization of morphological and biochemical features as well as advanced methods for data treatment and interrogation. Currently proposed MSI workflows are subject to several limitations, including nonoptimized raw data preprocessing, imprecise image coregistration, and limited pattern recognition capabilities. Here we outline a comprehensive strategy for histology-driven MSI, using desorption electrospray ionization that covers (i) optimized data preprocessing for improved information recovery; (ii) precise image coregistration; and (iii) efficient extraction of tissue-specific molecular ion signatures for enhanced biochemical distinction of different tissue types. The proposed workflow has been used to investigate region-specific lipid signatures in colorectal cancer tissue. Unique lipid patterns were observed using this approach according to tissue type, and a tissue recognition system using multivariate molecular ion patterns allowed highly accurate (>98%) identification of pixels according to morphology (cancer, healthy mucosa, smooth muscle, and microvasculature). This strategy offers unique insights into tumor microenvironmental biochemistry and should facilitate compilation of a large-scale tissue morphology-specific MSI spectral database with which to pursue next-generation, fully automated histological approaches.

Spatially resolved metabolic phenotyping of breast cancer by desorption electrospray ionization mass spectrometry.

Breast cancer is a heterogeneous disease characterized by varying responses to therapeutic agents and significant differences in long-term survival. Thus, there remains an unmet need for early diagnostic and prognostic tools and improved histologic characterization for more accurate disease stratification and personalized therapeutic intervention. This study evaluated a comprehensive metabolic phenotyping method in breast cancer tissue that uses desorption electrospray ionization mass spectrometry imaging (DESI MSI), both as a novel diagnostic tool and as a method to further characterize metabolic changes in breast cancer tissue and the tumor microenvironment. In this prospective single-center study, 126 intraoperative tissue biopsies from tumor and tumor bed from 50 patients undergoing surgical resections were subject to DESI MSI. Global DESI MSI models were able to distinguish adipose, stromal, and glandular tissue based on their metabolomic fingerprint. Tumor tissue and tumor-associated stroma showed evident changes in their fatty acid and phospholipid composition compared with normal glandular and stromal tissue. Diagnosis of breast cancer was achieved with an accuracy of 98.2% based on DESI MSI data (PPV 0.96, NVP 1, specificity 0.96, sensitivity 1). In the tumor group, correlation between metabolomic profile and tumor grade/hormone receptor status was found. Overall classification accuracy was 87.7% (PPV 0.92, NPV 0.9, specificity 0.9, sensitivity 0.92). These results demonstrate that DESI MSI may be a valuable tool in the improved diagnosis of breast cancer in the future. The identified tumor-associated metabolic changes support theories of de novo lipogenesis in tumor tissue and the role of stroma tissue in tumor growth and development and overall disease prognosis.

Multicenter Matrix-Assisted Laser Desorption/Ionization Mass Spectrometry Imaging (MALDI MSI) Identifies Proteomic Differences in Breast-Cancer-Associated Stroma

MALDI mass spectrometry imaging (MSI) has rapidly established itself as a powerful biomarker discovery tool. To date, no formal investigation has assessed the center-to-center comparability of MALDI MSI experiments, an essential step for it to develop into a new diagnostic method. To test such capabilities, we have performed a multicenter study focused on biomarkers of stromal activation in breast cancer. MALDI MSI experiments were performed in two centers using independent tissue banks, infrastructure, methods, and practitioners. One of the data sets was used for discovery and the other for validation. Areas of intra- and extratumoral stroma were selected, and their protein signals were compared. Four protein signals were found to be significantly associated with tumor-associated stroma in the discovery data set measured in Munich. Three of these peaks were also detected in the independent validation data set measured in Leiden, all of which were also significantly associated with intratumoral stroma. Hierarchical clustering displayed 100% accuracy in the Munich MSI data set and 80.9% accuracy in the Leiden MSI data set. The association of one of the identified mass signals (PA28) with stromal activation was confirmed with immunohistochemistry performed on 20 breast tumors. Independent and international MALDI MSI investigations could identify validated biomarkers of stromal activation.

Mineralized soft-tissue structure and chemistry in a mummified hadrosaur from the Hell Creek Formation, North Dakota (USA)

An extremely well-preserved dinosaur (Cf. Edmontosaurus sp.) found in the Hell Creek Formation (Upper Cretaceous, North Dakota) retains soft-tissue replacement structures and associated organic compounds. Mineral cements precipitated in the skin apparently follow original cell boundaries, partially preserving epidermis microstructure. Infrared and electron microprobe images of ossified tendon clearly show preserved mineral zonation, with silica and trapped carbon dioxide forming thin linings on Haversian canals within apatite. Furthermore, Fourier transform infrared spectroscopy (FTIR) of materials recovered from the skin and terminal ungual phalanx suggests the presence of compounds containing amide groups. Amino acid composition analyses of the mineralized skin envelope clearly differ from the surrounding matrix; however, intact proteins could not be obtained using protein mass spectrometry. The presence of endogenously derived organics from the skin was further demonstrated by pyrolysis gas chromatography mass spectrometry (Py-GCMS), indicating survival and presence of macromolecules that were in part aliphatic (see the electronic supplementary material).

Characterization and Identification of Clinically Relevant Microorganisms Using Rapid Evaporative Ionization Mass Spectrometry

Rapid evaporative ionization mass spectrometry (REIMS) was investigated for its suitability as a general identification system for bacteria and fungi. Strains of 28 clinically relevant bacterial species were analyzed in negative ion mode, and corresponding data was subjected to unsupervised and supervised multivariate statistical analyses. The created supervised model yielded correct cross-validation results of 95.9%, 97.8%, and 100% on species, genus, and Gram-stain level, respectively. These results were not affected by the resolution of the mass spectral data. Blind identification tests were performed for strains cultured on different culture media and analyzed using different instrumental platforms which led to 97.8-100% correct identification. Seven different Escherichia coli strains were subjected to different culture conditions and were distinguishable with 88% accuracy. In addition, the technique proved suitable to distinguish five pathogenic Candida species with 98.8% accuracy without any further modification to the experimental workflow. These results prove that REIMS is sufficiently specific to serve as a culture condition-independent tool for the identification and characterization of microorganisms.

Rapid Evaporative Ionization Mass Spectrometry Imaging Platform for Direct Mapping from Bulk Tissue and Bacterial Growth Media

Rapid evaporative ionization mass spectrometry (REIMS) technology allows real time intraoperative tissue classification and the characterization and identification of microorganisms. In order to create spectral libraries for training the classification models, reference data need to be acquired in large quantities as classification accuracy generally improves as a function of number of training samples. In this study, we present an automated high-throughput method for collecting REIMS data from heterogeneous organic tissue. The underlying instrumentation consists of a 2D stage with an additional high-precision z-axis actuator that is equipped with an electrosurgical diathermy-based sampling probe. The approach was validated using samples of human liver with metastases and bacterial strains, cultured on solid medium, belonging to the species P. aeruginosa, B. subtilis, and S. aureus. For both sample types, spatially resolved spectral information was obtained that resulted in clearly distinguishable multivariate clustering between the healthy/cancerous liver tissues and between the bacterial species.

Identification of hydroxyapatite spherules provides new insight into subretinal pigment epithelial deposit formation in the aging eye

Significance Proteins and lipids accumulating in deposits external to the retinal pigment epithelium (RPE) represent a barrier to metabolic exchange between the retina and the choroidal capillaries. With time, these deposits can lead to age-related macular degeneration (AMD), the most common cause of blindness in the elderly in the developed world. It remains unclear how sub-RPE deposits are initiated and grow to clinically relevant features. Using a combination of high-resolution analytical techniques, we found that tiny hydroxyapatite (bone mineral) spherules with cholesterol-containing cores are present in all examined sub-RPE deposits, providing a scaffold to which proteins adhere. If the spherules are important in initiating sub-RPE deposit formation, this finding may provide attractive new approaches for early identification and treatment of AMD. Accumulation of protein- and lipid-containing deposits external to the retinal pigment epithelium (RPE) is common in the aging eye, and has long been viewed as the hallmark of age-related macular degeneration (AMD). The cause for the accumulation and retention of molecules in the sub-RPE space, however, remains an enigma. Here, we present fluorescence microscopy and X-ray diffraction evidence for the formation of small (0.5–20 μm in diameter), hollow, hydroxyapatite (HAP) spherules in Bruch’s membrane in human eyes. These spherules are distinct in form, placement, and staining from the well-known calcification of the elastin layer of the aging Bruch’s membrane. Secondary ion mass spectrometry (SIMS) imaging confirmed the presence of calcium phosphate in the spherules and identified cholesterol enrichment in their core. Using HAP-selective fluorescent dyes, we show that all types of sub-RPE deposits in the macula, as well as in the periphery, contain numerous HAP spherules. Immunohistochemical labeling for proteins characteristic of sub-RPE deposits, such as complement factor H, vitronectin, and amyloid beta, revealed that HAP spherules were coated with these proteins. HAP spherules were also found outside the sub-RPE deposits, ready to bind proteins at the RPE/choroid interface. Based on these results, we propose a novel mechanism for the growth, and possibly even the formation, of sub-RPE deposits, namely, that the deposit growth and formation begin with the deposition of insoluble HAP shells around naturally occurring, cholesterol-containing extracellular lipid droplets at the RPE/choroid interface; proteins and lipids then attach to these shells, initiating or supporting the growth of sub-RPE deposits.

Imaging mass spectrometry to visualize biomolecule distributions in mouse brain tissue following hemispheric cortical spreading depression

MALDI mass spectrometry can simultaneously measure hundreds of biomolecules directly from tissue. Using essentially the same technique but different sample preparation strategies, metabolites, lipids, peptides and proteins can be analyzed. Spatially correlated analysis, imaging MS, enables the distributions of these biomolecular ions to be simultaneously measured in tissues. A key advantage of imaging MS is that it can annotate tissues based on their MS profiles and thereby distinguish biomolecularly distinct regions even if they were unexpected or are not distinct using established histological and histochemical methods e.g. neuropeptide and metabolite changes following transient electrophysiological events such as cortical spreading depression (CSD), which are spreading events of massive neuronal and glial depolarisations that occur in one hemisphere of the brain and do not pass to the other hemisphere , enabling the contralateral hemisphere to act as an internal control. A proof-of-principle imaging MS study, including 2D and 3D datasets, revealed substantial metabolite and neuropeptide changes immediately following CSD events which were absent in the protein imaging datasets. The large high dimensionality 3D datasets make even rudimentary contralateral comparisons difficult to visualize. Instead non-negative matrix factorization (NNMF), a multivariate factorization tool that is adept at highlighting latent features, such as MS signatures associated with CSD events, was applied to the 3D datasets. NNMF confirmed that the protein dataset did not contain substantial contralateral differences, while these were present in the neuropeptide dataset.