This systematic review will use 5 electronic databases—PubMed, Embase, Web of Science, Scopus, and CENTRAL—from inception to the present as primary data sources. These databases were selected to ensure broad coverage of biomedical and omics-related literature. In addition, specialized metabolomics repositories such as MetaboLights and Metabolomics Workbench will be manually searched to identify relevant studies and supplementary datasets not indexed in the aforementioned databases. Both authors collaborated to determine relevant search terms, including keywords and MeSH (Medical Subject Headings) terms, and search queries were constructed using Boolean operators (“AND” and “OR”). Searches will be conducted from the inception of each database to the most recent available studies without applying search filters. No language restrictions will be applied during the search. In cases in which potentially relevant articles are published in languages other than English, professional and credible translation services will be used to translate the texts into English for inclusion and assessment. Unpublished studies identified at the time of the final search will be excluded. The detailed search strategy is shown in Multimedia Appendix 1. All retrieved references will be downloaded in RIS format and managed using EndNote (version 19; Clarivate Analytics), which will be used to combine records from all databases and remove duplicates.

The screening of eligible studies will be undertaken after the database searches are complete. All identified references will be managed using Microsoft Excel, where duplicates will be identified and removed. To ensure consistency and familiarity with the screening process, at least 5 randomly selected studies will be pilot-tested by the reviewers. The evaluation will occur in 2 stages. Initially, the 2 reviewers (MNMN and PSNMK) will independently evaluate the titles and abstracts of all references based on the established eligibility criteria. During the second stage, the same 2 reviewers will evaluate the complete texts of possibly suitable articles for inclusion. Disagreements or differences between the reviewers will be resolved through discussion to achieve consensus. This methodology aims to reduce bias and guarantee a transparent and methodical selection of studies. We will document the selection process in adequate detail to finalize a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram [16].

Data extraction will use a standardized Microsoft Excel template to record analytical platforms (NMR spectroscopy, GC-MS, and LC-MS); quantified metabolites (such as amino acids and lipids); associated metabolic pathways; and statistical associations with BP parameters, including SBP or diastolic BP values and diagnostic accuracy metrics. The template will systematically capture study characteristics, including design, population demographics, interventions, comparators, and outcomes related to metabolite-BP linkages. The 2 reviewers (MNMN and PSNMK) will independently extract data from the selected papers, and any disagreements will be resolved through consensus. If further clarification or information is required, the corresponding authors of the original studies will be contacted.

For randomized controlled trials, the risk-of-bias assessment will use version 2 of the Cochrane risk-of-bias tool for randomized trials; for nonrandomized studies, the Risk of Bias in Nonrandomized Studies of Interventions tool will be applied. For observational studies examining metabolomic exposures and hypertension outcomes, which represent most metabolomics research designs, we will apply the Risk of Bias in Nonrandomized Studies of Exposures (ROBINS-E) tool [17]. The ROBINS-E tool is specifically designed for cohort and case-control studies evaluating biological exposures such as metabolite concentrations, assessing 7 bias domains: confounding, participant selection, exposure classification, deviations from intended exposures, missing data, outcome measurement, and selective reporting. To address metabolomics-specific methodological challenges within the ROBINS-E framework, we will systematically evaluate (1) preanalytical factors, including sample collection protocols, storage conditions, freeze-thaw cycles, and metabolite stability considerations; (2) analytical quality control procedures, including use of pooled quality control samples, internal standards, coefficient of variation reporting, and batch effect correction methods; (3) metabolite identification confidence according to Metabolomics Standards Initiative criteria (levels 1-4); and (4) data processing rigor, including normalization, missing value handling, and multiple testing correction. The 2 reviewers (MNMN and PSNMK) will independently assess each study, and any differences in their assessments will be resolved through discussion. If essential information is unclear or unavailable in the published reports, additional details will be sought by contacting the study investigators. This rigorous approach ensures a transparent and systematic evaluation of study quality and potential sources of bias. The certainty of evidence for each outcome will be assessed using the Grading of Recommendations Assessment, Development, and Evaluation approach, and a summary table will be presented, including justifications for each assessment.

A meta-analysis will be conducted if 2 or more studies provide identical outcomes; otherwise, the results will be described narratively. For metabolomics studies, “identical outcomes” will be defined as the same metabolite confirmed through International Chemical Identifier Key [18] matching, ensuring consistent identification across studies regardless of naming conventions, with cross-referencing through PubChem [19], the Human Metabolome Database [20], the Kyoto Encyclopedia of Genes and Genomes [21], and the Reference Set of Metabolite Names [22] to resolve nomenclature discrepancies. Data analysis will use The Cochrane Collaboration’s Review Manager software (version 5.4) for aggregating results. Continuous outcomes will be computed as mean differences or standardized mean differences with 95% CIs, depending on the consistency of the measuring instruments used across studies. Dichotomous outcomes will be represented as odds ratios accompanied by 95% CIs. The I² statistic will assess heterogeneity, with values of 0% to 40% indicating low heterogeneity, 30% to 60% indicating moderate heterogeneity, 50% to 90% indicating substantial heterogeneity, and 75% to 100% indicating considerable heterogeneity. Analyses will use random-effects models to account for heterogeneity, particularly when pooling across different metabolomics methodologies.

Given the inherent variability in metabolomics platforms (NMR spectroscopy vs LC-MS vs GC-MS) and metabolite panels, meta-analyses will be stratified by analytical platform type, with platform-specific pooled estimates calculated separately before assessing cross-platform comparability using subgroup heterogeneity tests (Cochran Q statistic; interaction P<.10). When substantial between-platform heterogeneity exists (I²>75% or significant platform interaction), results will be presented separately by platform rather than pooled across methodologies. Studies will be documented for their analytical platform specifications (instrument type, column chemistry for LC-MS, and field strength for NMR spectroscopy), and these characteristics will inform quality weighting in meta-analyses. Sensitivity analyses will be performed by (1) excluding each analytical platform sequentially to assess platform-specific influence on pooled estimates; (2) comparing targeted vs untargeted metabolomics approaches; (3) restricting analyses to studies using validated, standardized protocols; (4) excluding studies with high risk of bias; and (5) restricting analyses to studies with Metabolomics Standards Initiative level 1 and 2 metabolite identifications. Meta-regression will be performed when at least 10 studies are available, with analytical platform type, sample matrix (serum vs plasma vs urine), and study quality as moderator variables to quantify their contribution to heterogeneity. Predefined subgroups will include age, sex, geographic region, analytical platform (NMR spectroscopy vs LC-MS vs GC-MS), sample type (serum vs plasma vs urine), metabolite identification confidence level (levels 1-2 vs 3-4), study design (cross-sectional vs prospective cohort), and risk-of-bias categories. Assessment of publication bias will be conducted using funnel plot symmetry and the Egger regression approach when a significant number of papers (generally ≥10) are present. For outcomes characterized by inadequate data, an I² value of more than 75%, significant platform-related heterogeneity, or fewer than 2 studies reporting identical metabolites, or when cross-platform differences preclude reliable pooling, findings will be synthesized narratively with results stratified by analytical platform and metabolite class. This comprehensive approach ensures that metabolomics-specific methodological variability is systematically addressed through predefined strategies, enhancing the reliability and interpretability of pooled estimates.

This protocol has been registered with PROSPERO (CRD420251016289). Any protocol amendments will be documented on the PROSPERO record, including the date, description, and rationale for each change. The review will adhere to the PRISMA 2020 statement [23] and PRISMA-P (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols) [24] guidelines and will be reported according to the Cochrane Handbook for Systematic Reviews of Interventions [25]. The PRISMA 2020 checklist has been provided (Checklist 1).

There is increasing recognition of the important role that metabolomics can play in advancing the understanding, detection, and management of hypertension, yet large-scale, systematic data on its application in hypertensive populations remain limited. Comprehensive studies investigating metabolomic signatures and their association with BP are critical for elucidating the underlying mechanisms of hypertension and informing the development of more precise diagnostic and therapeutic strategies.

A central hypothesis in this field is that a combination of genetic, environmental, and lifestyle factors, particularly in vulnerable populations such as those with lower socioeconomic status or coexisting metabolic disorders, may contribute to distinct metabolomic profiles associated with increased hypertension risk [26]. Additionally, variability in the accuracy and validity of different metabolomic techniques may result in inconsistent detection of relevant biomarkers, potentially leading to underestimation or overestimation of hypertension risk. The interplay of multiple contextual risk factors and methodological differences could jointly influence the observed associations between metabolomic markers and hypertension [27].

This systematic review and meta-analysis will synthesize the available literature to clarify the relationship between metabolomic profiles and hypertension, including the prevalence and distribution of key biomarkers; the influence of risk factors such as age, sex, ethnicity, and comorbidities; and the impact of different detection methods on study outcomes. By integrating data from diverse populations and analytical platforms, this review aims to provide a more nuanced and comprehensive understanding of the role of metabolomics in hypertension. The findings are expected to highlight that factors such as socioeconomic status, lifestyle, and methodological limitations in metabolite detection are key barriers to the effective clinical translation of metabolomics in hypertension. Furthermore, the review will provide a transferable framework for the application of metabolomics in hypertension research and practice.

A limitation of this review is that data on metabolomic profiles in hypertension are often derived from secondary analyses of clinical or epidemiological datasets, which may be subjected to incomplete reporting or variability in sample handling and analytical protocols. While such data reflect real-world clinical practice, they may also introduce bias or limit the generalizability of the findings. Implementing a model that requires prospective, standardized metabolomic assessments may not be feasible in all health care settings, particularly where resources or expertise are limited. It is important to recognize that the proposed framework is preliminary and should not be directly applied in clinical practice without further validation. Another limitation is the potential for selection bias, as studies may vary in the timing of sample collection relative to disease onset or treatment and may exclude individuals with acute or secondary forms of hypertension, thereby restricting the applicability of the findings to the broader hypertensive population. This review is also limited by the exclusion of participants with major comorbidities or secondary hypertension, potentially reducing generalizability to real-world hypertensive populations. This criterion was applied to isolate metabolomic signatures specific to primary hypertension, and future research including comorbid groups will be needed to assess external validity.

The findings from this systematic review will inform future research directions by identifying priority areas for investigation, including the need for longitudinal studies that incorporate serial metabolomic assessments to capture temporal changes across different stages of hypertension development and progression. Validation of consistently identified metabolite-hypertension associations in independent, prospective cohorts will be critical for establishing causality and assessing the temporal stability of metabolomic signatures. The review will also identify methodological gaps requiring standardization, such as sample collection protocols, analytical platform harmonization, and data processing procedures, which must be addressed before metabolomic findings can be translated into clinical applications.

The clinical integration of metabolomics into hypertension management will require a staged approach informed by the evidence synthesized in this review. On the basis of the strength and consistency of the findings, several translational pathways can be envisioned. If the review identifies metabolites with robust, replicated associations with hypertension across multiple studies and platforms, these could be incorporated into research-focused risk assessment tools or used to stratify participants in clinical trials of antihypertensive interventions. This would not require changes to clinical guidelines but would enhance research efficiency and mechanistic understanding. Metabolite panels showing consistent associations with hypertension could be evaluated in targeted validation studies comparing their additive predictive value beyond conventional risk factors. If validated metabolite panels demonstrate clinically meaningful improvements in risk prediction (eg, reclassification of 10%-15% of individuals from intermediate to high- or low-risk categories), they could be considered for incorporation into hypertension screening algorithms for high-risk populations, similar to how lipid panels are used in cardiovascular risk assessment.