Molecular classification of brain tumor biopsies using solid-state magic angle spinning proton magnetic resonance spectroscopy and robust classifiers

Magnetic resonance spectroscopy for the study of cns malignancies

Despite intensive research, brain tumors are amongst the malignancies with the worst prognosis; therefore, a prompt diagnosis and thoughtful assessment of the disease is required. The resistance of brain tumors to most forms of conventional therapy has led researchers to explore the underlying biology in search of new vulnerabilities and biomarkers. The unique metabolism of brain tumors represents one potential vulnerability and the basis for a system of classification. Profiling this aberrant metabolism requires a method to accurately measure and report differences in metabolite concentrations. Magnetic resonance-based techniques provide a framework for examining tumor tissue and the evolution of disease. Nuclear Magnetic Resonance (NMR) analysis of biofluids collected from patients suffering from brain cancer can provide biological information about disease status. In particular, urine and plasma can serve to monitor the evolution of disease through the changes observed in the metabolic profiles. Moreover, cerebrospinal fluid can be utilized as a direct reporter of cerebral activity since it carries the chemicals exchanged with the brain tissue and the tumor mass. Metabolic reprogramming has recently been included as one of the hallmarks of cancer. Accordingly, the metabolic rewiring experienced by these tumors to sustain rapid growth and proliferation can also serve as a potential therapeutic target. The combination of 13C tracing approaches with the utilization of different NMR spectral modalities has allowed investigations of the upregulation of glycolysis in the aggressive forms of brain tumors, including glioblastomas, and the discovery of the utilization of acetate as an alternative cellular fuel in brain metastasis and gliomas. One of the major contributions of magnetic resonance to the assessment of brain tumors has been the non-invasive determination of 2-hydroxyglutarate (2HG) in tumors harboring a mutation in isocitrate dehydrogenase 1 (IDH1). The mutational status of this enzyme already serves as a key feature in the clinical classification of brain neoplasia in routine clinical practice and pilot studies have established the use of in vivo magnetic resonance spectroscopy (MRS) for monitoring disease progression and treatment response in IDH mutant gliomas. However, the development of bespoke methods for 2HG detection by MRS has been required, and this has prevented the wider implementation of MRS methodology into the clinic. One of the main challenges for improving the management of the disease is to obtain an accurate insight into the response to treatment, so that the patient can be promptly diverted into a new therapy if resistant or maintained on the original therapy if responsive. The implementation of 13C hyperpolarized magnetic resonance spectroscopic imaging (MRSI) has allowed detection of changes in tumor metabolism associated with a treatment, and as such has been revealed as a remarkable tool for monitoring response to therapeutic strategies. In summary, the application of magnetic resonance-based methodologies to the diagnosis and management of brain tumor patients, in addition to its utilization in the investigation of its tumor-associated metabolic rewiring, is helping to unravel the biological basis of malignancies of the central nervous system.

Pilot Multi-Omic Analysis of Human Bile from Benign and Malignant Biliary Strictures: A Machine-Learning Approach

Cholangiocarcinoma (CCA) and pancreatic adenocarcinoma (PDAC) may lead to the development of extrahepatic obstructive cholestasis. However, biliary stenoses can also be caused by benign conditions, and the identification of their etiology still remains a clinical challenge. We performed metabolomic and proteomic analyses of bile from patients with benign (n = 36) and malignant conditions, CCA (n = 36) or PDAC (n = 57), undergoing endoscopic retrograde cholangiopancreatography with the aim of characterizing bile composition in biliopancreatic disease and identifying biomarkers for the differential diagnosis of biliary strictures. Comprehensive analyses of lipids, bile acids and small molecules were carried out using mass spectrometry (MS) and nuclear magnetic resonance spectroscopy (¹H-NMR) in all patients. MS analysis of bile proteome was performed in five patients per group. We implemented artificial intelligence tools for the selection of biomarkers and algorithms with predictive capacity. Our machine-learning pipeline included the generation of synthetic data with properties of real data, the selection of potential biomarkers (metabolites or proteins) and their analysis with neural networks (NN). Selected biomarkers were then validated with real data. We identified panels of lipids (n = 10) and proteins (n = 5) that when analyzed with NN algorithms discriminated between patients with and without cancer with an unprecedented accuracy.

Applications of high-resolution magic angle spinning MRS in biomedical studies II-Human diseases

High-resolution magic angle spinning (HRMAS) MRS is a powerful method for gaining insight into the physiological and pathological processes of cellular metabolism. Given its ability to obtain high-resolution spectra of non-liquid biological samples, while preserving tissue architecture for subsequent histopathological analysis, the technique has become invaluable for biochemical and biomedical studies. Using HRMAS MRS, alterations in measured metabolites, metabolic ratios, and metabolomic profiles present the possibility to improve identification and prognostication of various diseases and decipher the metabolomic impact of drug therapies. In this review, we evaluate HRMAS MRS results on human tissue specimens from malignancies and non-localized diseases reported in the literature since the inception of the technique in 1996. We present the diverse applications of the technique in understanding pathological processes of different anatomical origins, correlations with in vivo imaging, effectiveness of therapies, and progress in the HRMAS methodology.

Exploration of human brain tumour metabolism using pairwise metabolite-metabolite correlation analysis (MMCA) of HR-MAS 1H NMR spectra

METHODS

We quantified 378 HRMAS 1H NMR spectra of human brain tumours (132 glioblastomas, 101 astrocytomas, 75 meningiomas, 37 oligodendrogliomas and 33 metastases) from the eTumour database and looked for metabolic interactions by metabolite-metabolite correlation analysis (MMCA).

RESULTS

All tumour types showed remarkably similar metabolic correlations. Lactate correlated positively with alanine, glutamate with glutamine; creatine + phosphocreatine (tCr) correlated positively with lactate, alanine and choline + phosphocholine + glycerophosphocholine (tCho), and tCho correlated positively with lactate; fatty acids correlated negatively with lactate, glutamate + glutamine (tGlut), tCr and tCho. Oligodendrogliomas had fewer correlations but they still fitted that pattern.

CONCLUSIONS

Possible explanations include (i) glycolytic tumour cells (the Warburg effect) generating pyruvate which is converted to lactate, alanine, glutamate and then glutamine; (ii) an association between elevated glycolysis and increased choline turnover in membranes; (iii) an increase in the tCr pool to facilitate phosphocreatine-driven glutamate uptake; (iv) lipid signals come from cytosolic lipid droplets in necrotic or pre-necrotic tumour tissue that has lower concentrations of anabolic and catabolic metabolites. Additional metabolite exchanges with host cells may also be involved. If tumours co-opt a standard set of biochemical mechanisms to grow in the brain, then drugs might be developed to disrupt those mechanisms.

Minimization of spectral pattern changes during HRMAS experiments at 37 degrees celsius by prior focused microwave irradiation

OBJECT

High-resolution magic angle spinning (HRMAS) magnetic resonance spectroscopy provides detailed metabolomic information from intact tissue. However, long acquisition times and high rotation speed may lead to timedependent spectral pattern changes, which may affect proper interpretation of results. We report a strategy to minimize those changes, even at physiological recording temperature.

MATERIALS AND METHODS

Glioblastoma(Gbm) tumours were induced in 12 mice by stereotactic injection of GL261 cells. Animals were sacrificed and tumours were removed and stored in liquid N2. Half of the samples were exposed to focused microwave (FMW) irradiation prior to HRMAS while the other half was not. Time-course experiments (374 min at 37°C, 9.4T, 3,000 Hz spinning rate) were carried out to monitor spectral pattern changes. Differences were assessed with Unianova test while post-HRMAS histopathology analysis was performed to assess tissue integrity.

RESULTS

Significant changes (up to 1.7 fold) were observed in samples without FMW irradiation in several spectral regions e.g. mobile lipids/lactate (0.90-1.30 ppm), acetate (1.90 ppm), N-acetyl aspartate (2.00 ppm), and Choline-containing compounds (3.19-3.25 ppm). No significant changes in the spectral pattern of FMW-irradiated samples were recorded.

CONCLUSION

We describe here a successful strategy to minimize spectral pattern changes in mouse Gbm samples using a FMW irradiation system.

Identifying malignant transformations in recurrent low grade gliomas using high resolution magic angle spinning spectroscopy

OBJECTIVE

The objective of this study was to determine whether metabolic parameters derived from ex vivo analysis of tissue samples are predictive of biologic characteristics of recurrent low grade gliomas (LGGs). This was achieved by exploring the use of multivariate pattern recognition methods to generate statistical models of the metabolic characteristics of recurrent LGGs that correlate with aggressive biology and poor clinical outcome.

METHODS

Statistical models were constructed to distinguish between patients with recurrent gliomas that had undergone malignant transformation to a higher grade and those that remained grade 2. The pattern recognition methods explored in this paper include three filter-based feature selection methods (chi-square, gain ratio, and two-way conditional probability), a genetic search wrapper-based feature subset selection algorithm, and five classification algorithms (linear discriminant analysis, logistic regression, functional trees, support vector machines, and decision stump logit boost). The accuracy of each pattern recognition framework was evaluated using leave-one-out cross-validation and bootstrapping.

MATERIALS

The population studied included fifty-three patients with recurrent grade 2 gliomas. Among these patients, seven had tumors that transformed to grade 4, twenty-four had tumors that transformed to grade 3, and twenty-two had tumors that remained grade 2. Image-guided tissue samples were obtained from these patients using surgical navigation software. Part of each tissue sample was examined by a pathologist for histological features and for consistency with the tumor grade diagnosis. The other part of the tissue sample was analyzed with ex vivo nuclear magnetic resonance (NMR) spectroscopy.

RESULTS

Distinguishing between recurrent low grade gliomas that transformed to a higher grade and those that remained grade 2 was achieved with 96% accuracy, using areas of the ex vivo NMR spectrum corresponding to myoinositol, 2-hydroxyglutarate, hypo-taurine, choline, glycerophosphocholine, phosphocholine, glutathione, and lipid. Logistic regression and decision stump boosting models were able to distinguish between recurrent gliomas that transformed to a higher grade and those that did not with 100% training accuracy (95% confidence interval [93-100%]), 96% leave-one-out cross-validation accuracy (95% confidence interval [87-100%]), and 96% bootstrapping accuracy (95% confidence interval [95-97%]). Linear discriminant analysis, functional trees, and support vector machines were able to achieve leave-one-out cross-validation accuracy above 90% and bootstrapping accuracy above 85%. The three feature ranking methods were comparable in performance.

CONCLUSIONS

This study demonstrates the feasibility of using quantitative pattern recognition methods for the analysis of metabolic data from brain tissue obtained during the surgical resection of gliomas. All pattern recognition techniques provided good diagnostic accuracies, though logistic regression and decision stump boosting slightly outperform the other classifiers. These methods identified biomarkers that can be used to detect malignant transformations in individual low grade gliomas, and can lead to a timely change in treatment for each patient.

Correlation chemical shift imaging with low-power adiabatic pulses and constant-density spiral trajectories

In this work we introduce the concept of correlation chemical shift imaging (CCSI). Novel CCSI pulse sequences are demonstrated on clinical scanners for two-dimensional Correlation Spectroscopy (COSY) and Total Correlation Spectroscopy (TOCSY) imaging experiments. To date there has been limited progress reported towards a feasible and robust multivoxel 2D COSY. Localized 2D TOCSY imaging is shown for the first time in this work. Excitation with adiabatic GOIA-W(16,4) pulses (Gradient Offset Independent Adiabaticity Wurst modulation) provides minimal chemical shift displacement error, reduced lipid contamination from subcutaneous fat, uniform optimal flip angles, and efficient mixing for coupled spins, while enabling short repetition times due to low power requirements. Constant-density spiral readout trajectories are used to acquire simultaneously two spatial dimensions and f(2) frequency dimension in (k(x),k(y),t(2)) space in order to speed up data collection, while f(1) frequency dimension is encoded by consecutive time increments of t(1) in (k(x),k(y),t(1),t(2)) space. The efficient spiral sampling of the k-space enables the acquisition of a single-slice 2D COSY dataset with an 8 × 8 matrix in 8:32  min on 3 T clinical scanners, which makes it feasible for in vivo studies on human subjects. Here we present the first results obtained on phantoms, human volunteers and patients with brain tumors. The patient data obtained by us represent the first clinical demonstration of a feasible and robust multivoxel 2D COSY. Compared to the 2D J-resolved method, 2D COSY and TOCSY provide increased spectral dispersion which scales up with increasing main magnetic field strength and may have improved ability to unambiguously identify overlapping metabolites. It is expected that the new developments presented in this work will facilitate in vivo application of 2D chemical shift correlation MRS in basic science and clinical studies.

Low-power adiabatic sequences for in vivo localized two-dimensional chemical shift correlated MR spectroscopy

Novel low-power adiabatic sequences are demonstrated for in vivo localized two-dimensional correlated MR spectroscopy, such as correlated spectroscopy and total correlated spectroscopy. The design is based on three new elements for in vivo two-dimensional MRS: the use of gradient modulated constant adiabaticity GOIA-W(16,4) pulses for (i) localization (correlated spectroscopy and total correlated spectroscopy) and (ii) mixing (total correlated spectroscopy), and (iii) the use of longitudinal mixing (z-filter) for magnetization transfer during total correlated spectroscopy. GOIA-W(16,4) provides accurate signal localization, and more importantly, lowers the SAR for both total correlated spectroscopy mixing and localization. Longitudinal mixing improves considerably (fivefolds) the efficiency of total correlated spectroscopy transfer. These are markedly different from previous 1D editing total correlated spectroscopy sequences using spatially nonselective pulses and transverse mixing. Fully adiabatic (adiabatic mixing with adiabatic localization) and semiadiabatic (adiabatic mixing with nonadiabatic localization) methods for two-dimensional total correlated spectroscopy are compared. Results are presented for simulations, phantoms, and in vivo two-dimensional spectra from healthy volunteers and patients with brain tumors obtained on 3T clinical platforms equipped with standard hardware. To the best of our knowledge, this is the first demonstration of in vivo adiabatic two-dimensional total correlated spectroscopy and fully adiabatic two-dimensional correlated spectroscopy. It is expected that these methodological developments will advance the in vivo applicability of multi(spectrally)dimensional MRS to reliably identify metabolic biomarkers.