Integrating Machine Learning in Metabolomics: A Path to Enhanced Diagnostics and Data Interpretation.

Metabolomics, leveraging techniques like NMR and MS, is crucial for understanding biochemical processes in pathophysiological states. This field, however, faces challenges in metabolite sensitivity, data complexity, and omics data integration. Recent machine learning advancements have enhanced data analysis and disease classification in metabolomics. This study explores machine learning integration with metabolomics to improve metabolite identification, data efficiency, and diagnostic methods. Using deep learning and traditional machine learning, it presents advancements in metabolic data analysis, including novel algorithms for accurate peak identification, robust disease classification from metabolic profiles, and improved metabolite annotation. It also highlights multiomics integration, demonstrating machine learning's potential in elucidating biological phenomena and advancing disease diagnostics. This work contributes significantly to metabolomics by merging it with machine learning, offering innovative solutions to analytical challenges and setting new standards for omics data analysis.

Fast Label-Free Metabolic Profile Recognition Identifies Phenylketonuria and Subtypes.

Phenylketonuria (PKU) is the most common inherited metabolic disease in humans. Clinical screening of newborn heel blood samples for PKU is costly and time-consuming because it requires multiple procedures, like isotope labeling and derivatization, and PKU subtype identification requires an additional urine sample. Delayed diagnosis of PKU, or subtype identification can result in mental disability. Here, plasmonic silver nanoshells are used for laser desorption/ionization mass spectrometry (MS) detection of PKU with label-free assay by recognizing metabolic profile in dried blood spot (DBS) samples. A total of 1100 subjects are recruited and each DBS sample can be processed in seconds. This platform achieves PKU screening with a sensitivity of 0.985 and specificity of 0.995, which is comparable to existing clinical liquid chromatography MS (LC-MS) methods. This method can process 360 samples per hour, compared with the LC-MS method which processes only 30 samples per hour. Moreover, this assay enables precise identification of PKU subtypes without the need for a urine sample. It is demonstrated that this platform enables high-performance and fast, low-cost PKU screening and subtype identification. This approach might be suitable for the detection of other clinically relevant biomarkers in blood or other clinical samples.

Identification and validation of serum metabolite biomarkers for endometrial cancer diagnosis.

Endometrial cancer (EC) stands as the most prevalent gynecological tumor in women worldwide. Notably, differentiation diagnosis of abnormity detected by ultrasound findings (e.g., thickened endometrium or mass in the uterine cavity) is essential and remains challenging in clinical practice. Herein, we identified a metabolic biomarker panel for differentiation diagnosis of EC using machine learning of high-performance serum metabolic fingerprints (SMFs) and validated the biological function. We first recorded the high-performance SMFs of 191 EC and 204 Non-EC subjects via particle-enhanced laser desorption/ionization mass spectrometry (PELDI-MS). Then, we achieved an area-under-the-curve (AUC) of 0.957-0.968 for EC diagnosis through machine learning of high-performance SMFs, outperforming the clinical biomarker of cancer antigen 125 (CA-125, AUC of 0.610-0.684, p < 0.05). Finally, we identified a metabolic biomarker panel of glutamine, glucose, and cholesterol linoleate with an AUC of 0.901-0.902 and validated the biological function in vitro. Therefore, our work would facilitate the development of novel diagnostic biomarkers for EC in clinics.

Deciphering the metabolic heterogeneity of hematopoietic stem cells with single-cell resolution.

Metabolic status is crucial for stem cell functions; however, the metabolic heterogeneity of endogenous stem cells has never been directly assessed. Here, we develop a platform for high-throughput single-cell metabolomics (hi-scMet) of hematopoietic stem cells (HSCs). By combining flow cytometric isolation and nanoparticle-enhanced laser desorption/ionization mass spectrometry, we routinely detected >100 features from single cells. We mapped the single-cell metabolomes of all hematopoietic cell populations and HSC subpopulations with different division times, detecting 33 features whose levels exhibited trending changes during HSC proliferation. We found progressive activation of the oxidative pentose phosphate pathway (OxiPPP) from dormant to active HSCs. Genetic or pharmacological interference with OxiPPP increased reactive oxygen species level in HSCs, reducing HSC self-renewal upon oxidative stress. Together, our work uncovers the metabolic dynamics during HSC proliferation, reveals a role of OxiPPP for HSC activation, and illustrates the utility of hi-scMet in dissecting metabolic heterogeneity of immunophenotypically defined cell populations.

Early Diagnosis and Prognosis Prediction of Pancreatic Cancer Using Engineered Hybrid Core-Shells in Laser Desorption/Ionization Mass Spectrometry.

Effective detection of bio-molecules relies on the precise design and preparation of materials, particularly in laser desorption/ionization mass spectrometry (LDI-MS). Despite significant advancements in substrate materials, the performance of single-structured substrates remains suboptimal for LDI-MS analysis of complex systems. Herein, designer Au@SiO2@ZrO2 core-shell substrates are developed for LDI-MS-based early diagnosis and prognosis of pancreatic cancer (PC). Through controlling Au core size and ZrO2 shell crystallization, signal amplification of metabolites up to 3 orders is not only achieved, but also the synergistic mechanism of the LDI process is revealed. The optimized Au@SiO2@ZrO2 enables a direct record of serum metabolic fingerprints (SMFs) by LDI-MS. Subsequently, SMFs are employed to distinguish early PC (stage I/II) from controls, with an accuracy of 92%. Moreover, a prognostic prediction scoring system is established with enhanced efficacy in predicting PC survival compared to CA19-9 (p < 0.05). This work contributes to material-based cancer diagnosis and prognosis.

Self-Assembled Hyperbranched Gold Nanoarrays Decode Serum United Urine Metabolic Fingerprints for Kidney Tumor Diagnosis.

Serum united urine metabolic analysis comprehensively reveals the disease status for kidney diseases in particular. Thus, the precise and convenient acquisition of metabolic molecular information from united biofluids is vitally important for clinical disease diagnosis and biomarker discovery. Laser desorption/ionization mass spectrometry (LDI-MS) presents various advantages in metabolic analysis; however, there remain challenges in ionization efficiency and MS signal reproducibility. Herein, we constructed a self-assembled hyperbranched black gold nanoarray (HyBrAuNA) assisted LDI-MS platform to profile serum united urine metabolic fingerprints (S-UMFs) for diagnosis of early stage renal cell carcinoma (RCC). The closely packed HyBrAuNA afforded strong electromagnetic field enhancement and high photothermal conversion efficacy, enabling effective ionization of low abundant metabolites for S-UMF collection. With a uniform nanoarray, the platform presented excellent reproducibility to ensure the accuracy of S-UMFs obtained in seconds. When it was combined with automated machine learning analysis of S-UMFs, early stage RCC patients were discriminated from the healthy controls with an area under the curve (AUC) > 0.99. Furthermore, we screened out a panel of 9 metabolites (4 from serum and 5 from urine) and related pathways toward early stage kidney tumor. In view of its high-throughput, fast analytical speed, and low sample consumption, our platform possesses potential in metabolic profiling of united biofluids for disease diagnosis and pathogenic mechanism exploration.

Identifying High Gleason Score Prostate Cancer by Prostate Fluid Metabolic Fingerprint-Based Multi-Modal Recognition.

Prostate cancer (PCa) is the second most common cancer in males worldwide. The Gleason scoring system, which classifies the pathological growth pattern of cancer, is considered one of the most important prognostic factors for PCa. Compared to indolent PCa, PCa with high Gleason score (h-GS PCa, GS ≥ 8) has greater clinical significance due to its high aggressiveness and poor prognosis. It is crucial to establish a rapid, non-invasive diagnostic modality to decipher patients with h-GS PCa as early as possible. In this study, ferric nanoparticle-assisted laser desorption/ionization mass spectrometry (FeNPALDI-MS) to extract prostate fluid metabolic fingerprint (PSF-MF) is employed and combined with the clinical features of patients, such as prostate-specific antigen (PSA), to establish a multi-modal diagnosis assisted by machine learning. This approach yields an impressive area under the curve (AUC) of 0.87 to diagnose patients with h-GS, surpassing the results of single-modal diagnosis using only PSF-MF or PSA, respectively. Additionally, using various screening methods, six key metabolites that exhibit greater diagnostic efficacy (AUC = 0.96) are identified. These findings also provide insights into related metabolic pathways, which may provide valuable information for further elucidation of the pathological mechanisms underlying h-GS PCa.

Diagnosis of Esophageal Squamous Cell Carcinoma by High-Performance Serum Metabolic Fingerprints: A Retrospective Study.

Esophageal squamous cell carcinoma (ESCC) is a highly prevalent and aggressive malignancy, and timely diagnosis of ESCC contributes to an increased cancer survival rate. However, current detection methods for ESCC mainly rely on endoscopic examination, limited by a relatively low participation rate. Herein, ferric-particle-enhanced laser desorption/ionization mass spectrometry (FPELDI MS) is utilized to record the serum metabolic fingerprints (SMFs) from a retrospective cohort (523 non-ESCC participants and 462 ESCC patients) to build diagnostic models toward ESCC. The PFELDI MS achieved high speed (≈30 s per sample), desirable reproducibility (coefficients of variation < 15%), and high throughput (985 samples with ≈124 200 data points for each spectrum). Desirable diagnostic performance with area-under-the-curves (AUCs) of 0.925-0.966 is obtained through machine learning of SMFs. Further, a metabolic biomarker panel is constructed, exhibiting superior diagnostic sensitivity (72.2-79.4%, p < 0.05) as compared with clinical protein biomarker tests (4.3-22.9%). Notably, the biomarker panel afforded an AUC of 0.844 (95% confidence interval [CI]: 0.806-0.880) toward early ESCC diagnosis. This work highlighted the potential of metabolic analysis for accurate screening and early detection of ESCC and offered insights into the metabolic characterization of diseases including but not limited to ESCC.

Comprehensive Metabolic Fingerprints Characterize Neuromyelitis Optica Spectrum Disorder by Nanoparticle-Enhanced Laser Desorption/Ionization Mass Spectrometry.

Timely screening of neuromyelitis optica spectrum disorder (NMOSD) and differential diagnosis from myelin oligodendrocyte glycoprotein associated disorder (MOGAD) are the keys to improving the quality of life of patients. Metabolic disturbance occurs with the development of NMOSD. Still, advanced tools are required to probe the metabolic phenotype of NMOSD. Here, we developed a fast nanoparticle-enhanced laser desorption/ionization mass spectrometry assay for multiplexing metabolic fingerprints (MFs) from trace plasma and cerebrospinal fluid (CSF) samples in 30 s. Machine learning of the plasma MFs achieved the timely screening of NMOSD from healthy donors with an area under receiver operator characteristic curve (AUROC) of 0.998, and it comprehensively revealed the dysregulated neurotransmitter and energy metabolisms. Combining comprehensive MFs from both plasma and CSF, we constructed an integrated panel for differential diagnosis of NMOSD versus MOGAD with an AUROC of 0.923. This approach demonstrated performance superior to that of human experts in classifying two diseases, especially in antibody assay-limited regions. Together, this approach provides an advanced nanomaterial-based tool for identifying vulnerable populations below the antibody threshold of aquaporin-4 positivity.

Efficient plasma metabolic fingerprinting as a novel tool for diagnosis and prognosis of gastric cancer: a large-scale, multicentre study.

OBJECTIVE: Metabolic biomarkers are expected to decode the phenotype of gastric cancer (GC) and lead to high-performance blood tests towards GC diagnosis and prognosis. We attempted to develop diagnostic and prognostic models for GC based on plasma metabolic information. DESIGN: We conducted a large-scale, multicentre study comprising 1944 participants from 7 centres in retrospective cohort and 264 participants in prospective cohort. Discovery and verification phases of diagnostic and prognostic models were conducted in retrospective cohort through machine learning and Cox regression of plasma metabolic fingerprints (PMFs) obtained by nanoparticle-enhanced laser desorption/ionisation-mass spectrometry (NPELDI-MS). Furthermore, the developed diagnostic model was validated in prospective cohort by both NPELDI-MS and ultra-performance liquid chromatography-MS (UPLC-MS). RESULTS: We demonstrated the high throughput, desirable reproducibility and limited centre-specific effects of PMFs obtained through NPELDI-MS. In retrospective cohort, we achieved diagnostic performance with areas under curves (AUCs) of 0.862-0.988 in the discovery (n=1157 from 5 centres) and independent external verification dataset (n=787 from another 2 centres), through 5 different machine learning of PMFs, including neural network, ridge regression, lasso regression, support vector machine and random forest. Further, a metabolic panel consisting of 21 metabolites was constructed and identified for GC diagnosis with AUCs of 0.921-0.971 and 0.907-0.940 in the discovery and verification dataset, respectively. In the prospective study (n=264 from lead centre), both NPELDI-MS and UPLC-MS were applied to detect and validate the metabolic panel, and the diagnostic AUCs were 0.855-0.918 and 0.856-0.916, respectively. Moreover, we constructed a prognosis scoring system for GC in retrospective cohort, which can effectively predict the survival of GC patients. CONCLUSION: We developed and validated diagnostic and prognostic models for GC, which also contribute to advanced metabolic analysis towards diseases, including but not limited to GC.

Efficient Metabolic Fingerprinting of Follicular Fluid Encodes Ovarian Reserve and Fertility.

Ovarian reserve (OR) and fertility are critical in women's healthcare. Clinical methods for encoding OR and fertility rely on the combination of tests, which cannot serve as a multi-functional platform with limited information from specific biofluids. Herein, metabolic fingerprinting of follicular fluid (MFFF) from follicles is performed, using particle-assisted laser desorption/ionization mass spectrometry (PALDI-MS) to encode OR and fertility. PALDI-MS allows efficient MFFF, showing fast speed (≈30 s), high sensitivity (≈60 fmol), and desirable reproducibility (coefficients of variation <15%). Further, machine learning of MFFF is applied to diagnose diminished OR (area under the curve of 0.929) and identify high-quality oocytes/embryos (p < 0.05) by a single PALDI-MS test. Meanwhile, metabolic biomarkers from MFFF are identified, which also determine oocyte/embryo quality (p < 0.05) from the sampling follicles toward fertility prediction in clinics. This approach offers a powerful platform in women's healthcare, not limited to OR and fertility.

Diagnosis of Unruptured Intracranial Aneurysm by High-Performance Serum Metabolic Fingerprints.

Unruptured intracranial aneurysm (UIA) is a high-risk cerebrovascular saccular dilatation, the effective medical management of which depends on high-performance diagnosis. However, most UIAs are diagnosed incidentally during neurovascular imaging modalities, which are time-consuming and harmful (e.g., radiation). Serum metabolic fingerprints is a promising alternative for early diagnosis of UIA. Here, nanoparticle enhanced laser desorption/ionization mass spectrometry is applied to obtain high-performance UIA-specific serum metabolic fingerprints. Diagnostic performance with an area-under-the-curve (AUC) of 0.842 (95% confidence interval (CI): 0.783-0.891) is achieved by the constructed machine learning (ML) model, including ML algorithm selection and feature selection. Lactate, glutamine, homoarginine, and 3-methylglutaconic acid are identified as the metabolic biomarker panel, which showed satisfactory diagnosis (AUC of 0.812, 95% CI: 0.727-0.897) and effective growth risk assessment (p<0.05, two-tailed t-test) of UIAs. This work aims to promote the diagnostics of UIAs and metabolic biomarker screening for medical management.

Deep Learning of Dual Plasma Fingerprints for High-Performance Infection Classification.

Infection classification is the key for choosing the proper treatment plans. Early determination of the causative agents is critical for disease control. Host responses analysis can detect variform and sensitive host inflammatory responses to ascertain the presence and type of the infection. However, traditional host-derived inflammatory indicators are insufficient for clinical infection classification. Fingerprints-based omic analysis has attracted increasing attention globally for analyzing the complex host systemic immune response. A single type of fingerprints is not applicable for infection classification (area under curve (AUC) of 0.550-0.617). Herein, an infection classification platform based on deep learning of dual plasma fingerprints (DPFs-DL) is developed. The DPFs with high reproducibility (coefficient of variation <15%) are obtained at low sample consumption (550 nL native plasma) using inorganic nanoparticle and organic matrix assisted laser desorption/ionization mass spectrometry. A classifier (DPFs-DL) for viral versus bacterial infection discrimination (AUC of 0.775) and coronavirus disease 2019 (COVID-2019) diagnosis (AUC of 0.917) is also built. Furthermore, a metabolic biomarker panel of two differentially regulated metabolites, which may serve as potential biomarkers for COVID-19 management (AUC of 0.677-0.883), is constructed. This study will contribute to the development of precision clinical care for infectious diseases.

A Copper-Based Biosensor for Dual-Mode Glucose Detection.

Glucose is a source of energy for daily activities of the human body and is regarded as a clinical biomarker, due to the abnormal glucose level in the blood leading to many endocrine metabolic diseases. Thus, it is indispensable to develop simple, accurate, and sensitive methods for glucose detection. However, the current methods mainly depend on natural enzymes, which are unstable, hard to prepare, and expensive, limiting the extensive applications in clinics. Herein, we propose a dual-mode Cu2O nanoparticles (NPs) based biosensor for glucose analysis based on colorimetric assay and laser desorption/ionization mass spectrometry (LDI MS). Cu2O NPs exhibited excellent peroxidase-like activity and served as a matrix for LDI MS analysis, achieving visual and accurate quantitative analysis of glucose in serum. Our proposed method possesses promising application values in clinical disease diagnostics and monitoring.

Defective Fe Metal-Organic Frameworks Enhance Metabolic Profiling for High-Accuracy Diagnosis of Human Cancers.

Cancers heavily threaten human life; therefore, a high-accuracy diagnosis is vital to protect human beings from the suffering of cancers. While biopsies and imaging methods are widely used as current technologies for cancer diagnosis, a new detection platform by metabolic analysis is expected due to the significant advantages of fast, simple, and cost-effectiveness with high body tolerance. However, the signal of molecule biomarkers is too weak to acquire high-accuracy diagnosis. Herein, precisely engineered metal-organic frameworks for laser desorption/ionization mass spectrometry, allowing favorable charge transfer within the molecule-substrate interface and mitigated thermal dissipation by adjusting the phonon scattering with metal nodes, are developed. Consequently, a surprising signal enhancement of ≈10 000-fold is achieved, resulting in diagnosis of three major cancers (liver/lung/kidney cancer) with area-under-the-curve of 0.908-0.964 and accuracy of 83.2%-90.6%, which promises a universal detection tool for large-scale clinical diagnosis of human cancers.

Glaucoma Characterization by Machine Learning of Tear Metabolic Fingerprinting.

Glaucoma is a common optic neuropathy disease affecting over 76 million people. Both timely diagnosis and progression monitoring are critical but challenging. Conventional characterization of glaucoma needs a combination of methods, calling for tedious procedures and experienced doctors. Herein, a platform through machine learning of tear metabolic fingerprinting (TMF) using nanoparticle enhanced laser desorption-ionization mass spectrometry is built. Direct TMF is obtained noninvasively, with fast speed and high reproducibility, using trace tear samples (down to 10 nL). Consequently, glaucoma patients are screened against healthy controls with the area under the curve (AUC) of 0.866, through machine learning of TMF. Further, primary open-angle glaucoma (POAG) is differentiated from primary angle-closure glaucoma (PACG) and an early-stage POAG is identified. Finally, a biomarker panel of six metabolites for glaucoma characterization (including screening, subtyping, and early diagnosis) with AUC of 0.827-0.891 is constructed, showing related metabolic pathways. The work will provide insights into eye diseases not limited to glaucoma.

Diagnosis and prognosis of breast cancer by high-performance serum metabolic fingerprints.

High-performance metabolic analysis is emerging in the diagnosis and prognosis of breast cancer (BrCa). Still, advanced tools are in demand to deliver the application potentials of metabolic analysis. Here, we used fast nanoparticle-enhanced laser desorption/ionization mass spectrometry (NPELDI-MS) to record serum metabolic fingerprints (SMFs) of BrCa in seconds, achieving high reproducibility and low consumption of direct serum detection without treatment. Subsequently, machine learning of SMFs generated by NPELDI-MS functioned as an efficient readout to distinguish BrCa from non-BrCa with an area under the curve of 0.948. Furthermore, a metabolic prognosis scoring system was constructed using SMFs with effective prediction performance toward BrCa (P < 0.005). Finally, we identified a biomarker panel of seven metabolites that were differentially enriched in BrCa serum and their related pathways. Together, our findings provide an efficient serum metabolic tool to characterize BrCa and highlight certain metabolic signatures as potential diagnostic and prognostic factors of diseases including but not limited to BrCa.

Serum Metabolic Fingerprints on Bowl-Shaped Submicroreactor Chip for Chemotherapy Monitoring.

Chemotherapy is a primary cancer treatment strategy, the monitoring of which is critical to enhancing the survival rate and quality of life of cancer patients. However, current chemotherapy monitoring mainly relies on imaging tools with inefficient sensitivity and radiation invasiveness. Herein, we develop the bowl-shaped submicroreactor chip of Au-loaded 3-aminophenol formaldehyde resin (denoted as APF-bowl&Au) with a specifically designed structure and Au loading content. The obtained APF-bowl&Au, used as the matrix of laser desorption/ionization mass spectrometry (LDI MS), possesses an enhanced localized electromagnetic field for strengthened small metabolite detection. The APF-bowl&Au enables the extraction of serum metabolic fingerprints (SMFs), and machine learning of the SMFs achieves chemotherapy monitoring of ovarian cancer with area-under-the-curve (AUC) of 0.81-0.98. Furthermore, a serum metabolic biomarker panel is preliminarily identified, exhibiting gradual changes as the chemotherapy cycles proceed. This work provides insights into the development of nanochips and contributes to a universal detection platform for chemotherapy monitoring.

Anion-Afforded Functions of Ionic Metal-Organic Frameworks: Ionochromism, Anion Conduction, and Catalysis.

The exchangeable counterions in ionic metal-organic frameworks (IMOFs) provide facile and versatile handles to manipulate functions associated with the ionic guests themselves and host-guest interactions. However, anion-exchangeable stable IMOFs combining multiple anion-related functions are still undeveloped. In this work, a novel porous IMOF featuring unique self-penetration was constructed from an electron-deficient tris(pyridinium)-tricarboxylate zwitterionic ligand. The water-stable IMOF undergoes reversible and single-crystal-to-single-crystal anion exchange and shows selective and discriminative ionochromic behaviors toward electron-rich anions owing to donor-acceptor interactions. The IMOFs with different anions are good ionic conductors with low activation energy, the highest conductivity being observed with chloride. Furthermore, integrating Lewis acidic sites and nucleophilic guest anions in solid state, the IMOFs act as heterogeneous and recyclable catalysts to efficiently catalyze the cycloaddition of CO2 to epoxides without needing the use of halide cocatalysts. The catalytic activity is strongly dependent upon the guest anions, and the iodide shows the highest activity. The results demonstrate the great potential of developing IMOFs with various functions related to the guest ions included in the porous matrices.

Monitoring Retinoblastoma by Machine Learning of Aqueous Humor Metabolic Fingerprinting.

The most common intraocular pediatric malignancy, retinoblastoma (RB), accounts for ≈10% of cancer in children. Efficient monitoring can enhance living quality of patients and 5-year survival ratio of RB up to 95%. However, RB monitoring is still insufficient in regions with limited resources and the mortality may even reach over 70% in such areas. Here, an RB monitoring platform by machine learning of aqueous humor metabolic fingerprinting (AH-MF) is developed, using nanoparticle enhanced laser desorption/ionization mass spectrometry (LDI MS). The direct AH-MF of RB free of sample pre-treatment is recorded, with both high reproducibility (coefficient of variation < 10%) and sensitivity (low to 0.3 pmol) at sample volume down to 40 nL only. Further, early and advanced RB patients with area-under-the-curve over 0.9 and accuracy over 80% are differentiated, through machine learning of AH-MF. Finally, a metabolic biomarker panel of 7 metabolites through accurate MS and tandem MS (MS/MS) with pathway analysis to monitor RB is identified. This work can contribute to advanced metabolic analysis of eye diseases including but not limited to RB and screening of new potential metabolic targets toward therapeutic intervention.