Metabolomic Profiles in Down Syndrome: A Scoping Review of Convergent and Context‐Dependent Patterns

ABSTRACT

Background

Individuals with Down syndrome (DS) face a high burden of health complications, yet the molecular underpinnings remain incompletely defined.

Objective

The objective of this study is to systematically identify metabolomic changes in individuals with DS and how they relate to DS‐associated conditions.

Methods

A scoping review of the literature was performed across four online databases to identify studies profiling metabolites in people with DS using untargeted or targeted metabolomics procedures. The findings were narratively synthesised to provide a comprehensive overview of patterns of convergence and variability across studies.

Results

Thirty‐four studies examining metabolites in individuals with DS were identified. The combined findings revealed widespread disruptions in energy (e.g., tricarboxylic acid cycle intermediates and acylcarnitines), one‐carbon (e.g., methionine and the SAM/SAH axis), amino acid (e.g., tryptophan‐kynurenine and glutamate/GABA) and lipid (e.g., phospholipids and sphingolipids) metabolism, along with changes in immune and neurotransmitter pathways. These metabolic alterations are associated with phenotypic variability and comorbidities in DS. However, the evidence reflects partially convergent and context‐dependent patterns, with substantial variability across studies.

Conclusion

Metabolic disturbances are common in DS, suggesting candidate metabolic signatures that still require independent replication and validation. Current evidence is predominantly cross‐sectional and associative, limiting causal inference. Integrating metabolomics with multi‐omics approaches may enhance the understanding of DS‐related health issues and support future translation into clinical applications.

1. Introduction

Down syndrome (DS), a condition caused by the full or partial triplication of human chromosome 21 (HSA21), remains the most common chromosomal disorder in humans (Antonarakis et al. 2020; Bull 2020). In 2021, approximately 1.57 million individuals worldwide were living with this condition, with an estimated incidence of 1 in 700 to 1000 live births (GBD 2021; Chen et al. 2022). DS is associated with a wide range of health issues, including intellectual disability, leukaemia, thyroid problems, congenital heart defects, obstructive sleep apnoea and dementia (Bull 2020; Baksh et al. 2023). As life expectancy in individuals with DS has increased from about 9 years in 1929 to over 60 in recent years, there has been a rising prevalence of additional health problems, mainly visual and hearing impairments, obesity, osteoporosis and dementia (Penrose 1949; Bittles and Glasson 2010; Glasson et al. 2014; Hodapp et al. 2016). These epidemiological trends have motivated efforts to improve health monitoring and clinical care for this population (Lagan et al. 2020; Rojnueangnit et al. 2020; Bull et al. 2022).

To clarify the underlying mechanisms of DS‐associated comorbidities, numerous investigations have been conducted (Wiseman et al. 2009; Antonarakis 2017; Drachman et al. 2017; Zhang et al. 2019; Chen et al. 2021). A recent paradigm shift considers DS not only as a genetic disorder but also as a potentially treatable condition and even as a possible metabolic disease, highlighting the existence of modifiable biological pathways (McCabe and McCabe 2013; Dierssen et al. 2020). Historically, research has emphasised the overexpression of HSA21‐encoded genes to explain the development of various DS‐related phenotypes (Antonarakis 2017; Chapman et al. 2024). However, advances in omics sciences now provide systems‐level frameworks to interrogate the molecular basis of these complex traits.

Throughout the 21st century, omics sciences have become powerful tools for understanding the molecular mechanisms of disease (Wang et al. 2020; Babu and Snyder 2023). Within this landscape, metabolomics enables the simultaneous measurement of hundreds of metabolites from biological samples, offering a comprehensive snapshot of biochemical activity (Wishart et al. 2007). Because metabolomics readouts capture changes in both concentrations and metabolic fluxes, the metabolome provides a dynamic and integrative view of the physiological state at a given time (Lindon et al. 2004; Schmidt 2004; Kaddurah‐Daouk and Krishnan 2009). In trisomy 21, multiple pathophysiological processes, including oxidative stress, early‐onset neurodegeneration, immune dysregulation and mitochondrial dysfunction, are consistently reported and are thought to arise, at least in part, from gene‐dosage imbalances (Baksh et al. 2023). Nonetheless, these processes are typically studied in isolation, and their potential interconnections remain poorly understood. This fragmentation limits the ability to construct a coherent biological framework that explains how molecular alterations translate into the complex and heterogeneous phenotype observed in DS. Given that these processes are likely to share common metabolic determinants, metabolomics offers a particularly suitable approach to capture their interdependencies at a systems level (Pagano and Castello 2012; Illouz et al. 2021; Bhattacharyya et al. 2025; Rafii et al. 2025). However, findings remain dispersed across heterogeneous cohorts, analytical platforms and biological matrices, with limited efforts to synthesise patterns of convergence, directionality and biological relevance. Addressing this gap is essential to move beyond descriptive findings toward a more integrated understanding of metabolic dysregulation in DS. In this context, leveraging the integrative nature of metabolomics may provide a valuable framework to delineate genotype–phenotype relationships and identify biologically meaningful, potentially actionable metabolic pathways for targeted intervention.

Therefore, this scoping review aims to synthesise systems‐level metabolomic findings in individuals with DS across biofluids, tissues, age groups and analytical platforms, relate these alterations to DS‐associated conditions and provide an updated, methodologically transparent overview.

2. Methods

This review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) Extension for Scoping Reviews guidelines (Tricco et al. 2018).

2.1. Literature and Search Strategy

A comprehensive literature search was conducted across PubMed, Virtual Health Library, Web of Science and Embase to identify relevant studies published up to April 2025. In addition, a manual search of reference lists from eligible articles and forward citation tracking was performed to find records that the initial database search might have missed. Search strategy strings combined indexed terms, related keywords and synonyms for DS and metabolomics; minor source‐specific adjustments were applied to optimise retrieval (Table 1).

TABLE 1.

Search strategies' strings used per database.

Database	Search strategy
PubMed	(‘Down Syndrome’[Mesh] OR ‘Trisomy 21*’ OR ‘Syndrome, Down*’ OR ‘Down* Syndrome*’ OR Mongolism OR ‘Trisomy G’) AND (‘Metabolomics’[Mesh] OR ‘Metabolome’[Mesh] OR Metabolom* OR Metabonom* OR Metabolit* OR ‘Metabolic Profiling’ OR ‘Targeted Metabolomics’ OR ‘Targeted Metabonomics’ OR ‘Untargeted Metabolomics’ OR ‘Untargeted Metabonomics’)
Virtual Health Library	(mh:(‘Down Syndrome’) OR ‘Trisomy 21*’ OR ‘Syndrome, Down*’ OR ‘Down* Syndrome*’ OR Mongolism OR ‘Trisomy G’) AND (mh:(‘Metabolomics’) OR mh:(‘Metabolome’) OR Metabolom* OR Metabonom* OR Metabolit* OR ‘Metabolic Profiling’ OR ‘Targeted Metabolomics’ OR ‘Targeted Metabonomics’ OR ‘Untargeted Metabolomics’ OR ‘Untargeted Metabonomics’)
EMBASE	(‘down syndrome’/exp OR ‘trisomy 21’ OR ‘syndrome, down’ OR ‘down* syndrome*’ OR mongolism OR ‘trisomy g’) AND (‘metabolomics’/exp OR ‘metabolome’/exp OR metabolom* OR metabonom* OR metabolit* OR ‘metabolic profiling’ OR ‘targeted metabolomics’ OR ‘targeted metabonomics’ OR ‘untargeted metabolomics’ OR ‘untargeted metabonomics’)
Web of Science	(‘Down Syndrome’ OR ‘Trisomy 21*’ OR ‘Syndrome, Down*’ OR ‘Down* Syndrome*’ OR Mongolism OR ‘Trisomy G’) AND (‘Metabolomics’ OR ‘Metabolome’ OR Metabolom* OR Metabonom* OR Metabolit* OR ‘Metabolic Profiling’ OR ‘Targeted Metabolomics’ OR ‘Targeted Metabonomics’ OR ‘Untargeted Metabolomics’ OR ‘Untargeted Metabonomics’)

2.2. Eligibility Criteria

Studies were included based on meeting all of the following criteria: (1) original observational research; (2) participants included individuals with DS; (3) metabolomic analyses were performed on human‐derived biofluids or tissues; (4) metabolites were identified using mass spectrometry (MS), nuclear magnetic resonance (NMR) or magnetic resonance spectroscopy (MRS); and (5) results were reported separately for DS groups. Studies that did not meet these criteria, focused on cell or animal models or lacked metabolomics outcomes were excluded.

2.3. Screening Process

Records were imported onto the Rayyan platform (Ouzzani et al. 2016), where duplicates were identified and removed. Two reviewers independently screened the records' titles and abstracts against the inclusion criteria, resolved disagreements through discussion and consensus and then evaluated the full texts of potentially eligible studies. Studies that did not meet the inclusion criteria or lacked full‐text versions were excluded from the review.

2.4. Data Extraction and Synthesis of Evidence

Data from the final set of included studies were extracted into a structured CSV file by one reviewer and then verified by two additional reviewers, who discussed the findings with the authors. Extracted variables included: first author's name, year of publication, number of participants in the DS and control (if applicable) groups, participants' ages, sample type analysed, metabolomics platform and approach used, reported functional or pathway analyses, data availability and key findings. Recurrent findings were mapped to broader metabolic domains (e.g., energy, amino acid and lipid metabolism). Neither quantitative meta‐analysis nor pooling of effect sizes was attempted.

2.5. Appraisal of Methodological Context

Consistent with the aims of a scoping review, no formal risk‐of‐bias assessment was performed. However, study designs, cohort sizes, sample handling, analytical platforms and reporting of quality control procedures to contextualise the robustness and interpretability of the evidence were documented.

3. Results

3.1. Study Selection and Characteristics of Included Studies

A comprehensive overview of the study selection process is provided in the form of a PRISMA 2020 flow diagram (Figure 1). The database search yielded 1461 records. After removing duplicates, 925 unique records remained, of which 887 were excluded after title and abstract screening. Thirty‐eight studies underwent full‐text review; eight were excluded due to ineligible study design (no metabolomics per inclusion criteria), one for incorrect publication type and one for not focusing on individuals with DS. Additionally, six records identified through manual searching were included. In total, 34 studies met the eligibility criteria and were included in this review. The main characteristics of these studies are summarised in Table 2. Age reporting was highly variable between studies. Thirteen studies focused on children or adolescents, while four examined adults over 45 years old. Across all studies, participants with DS ranged from 6 months to 63 years old. Most studies included a euploid control group for comparison. These characteristics reflect a heterogeneous evidence base spanning multiple age groups and clinical contexts.

FIGURE 1.

PRISMA 2020 flow chart representation of the collection and screening process to retrieve relevant studies.

TABLE 2.

General characteristics of the included studies.

First author (year)	DS group (n)	CT group (n)	DS age		CT age		Metabolomic approach
			Mean ± STD	Min.–max.	Mean ± STD	Min.–max.
Murata et al. (1993)	18	15	33.2 ± 8.8	20.0–46.0	35.8 ± 8.4	—	Targeted
Shetty et al. (1995)	10	10	43.3 ± 15.1	22.0–63.0	47.5 ± 17.1	23.0–69.0	Targeted
Berry et al. (1999)	14	13	—	1.2–13.6	—	1.3–11.7	Targeted
Huang et al. (1999)	19	17	—	28.0–62.0	—	22.0–64.0	Targeted
Śmigielska‐Kuzia and Sobaniec (2007)	14	20	10.92 ± 3.49	6.0–17.0	11.78 ± 3.92	7.0–17.0	Targeted
Śmigielska‐Kuzia et al. (2010)	20	20	6.7 ± 3.72	3.0–15.0	10.9 ± 2.44	6.0–15.0	Targeted
Lamar et al. (2011)	18	13	47.2 ± 8.4	—	50.6 ± 10.4	—	Targeted
DS with dementia	17	13	52.8 ± 5.7	—	50.6 ± 10.4	—	Targeted
Biselli et al. (2012)	90	—	1.29 a	0.07–30.35	—	—	Targeted
Obeid et al. (2012)	35	47	11.0 a	1.9–27.0	13.0 a	5.7–17.0	Targeted
Tansley et al. (2012)	20	22	46.29	39.0–49.0	43.77	29.0–61.0	Untargeted
Mendes et al. (2013)	85	—	1.36 a	0.07–30.35	—	—	Targeted
Phillips et al. (2013)	14	11	32.7 ± 2.2 c	—	29.8 ± 2.7 c	—	Targeted
Tan et al. (2014)	35	39	35.0 ± 12.0	—	35.0 ± 12.0	—	Targeted
DS with dementia	11	39	52.0 ± 12.0	—	35.0 ± 12.0	—	Targeted
Lin et al. (2016)	22	15	48.3 ± 7.6	—	47.8 ± 7.6	—	Targeted
Culp‐Hill et al. (2017)	30	67	—	0.5–53.6	—	12.0–76.5	Untargeted
Caracausi et al. (2018)	67	29	11.3 ± 7.0	2.05–35.01	16.6 ± 7.8	3.66–31.55	Untargeted
Signorini et al. (2018)	16	61	24.9 ± 4.7	—	—	—	Targeted
Gross et al. (2019)	68	—	46.1 ± 1.3 c	—	—	—	Both
DS with AD	78	—	53.7 ± 0.7 c	—	—	—	Both
Powers et al. (2019)	199	194	28.1 ± 15.88	0.5–60.94	33.55 ± 13.89	0.6–56.1	Untargeted
Orozco et al. (2019)	31	193	3.7 ± 0.76	2.0–5.0	3.59 ± 0.81	2.0–5.0	Untargeted
Antonaros et al. (2020)	129	46	11.23 ± 6.64	2.05–36.96	15.18 ± 7.99	3.33–42.32	Untargeted
Mapstone et al. (2020)	211	—	42.16 ± 0.59 c	—	—	—	Untargeted
DS with AD	38	—	54.39 ± 0.93 c	—	—	—	Untargeted
DS with MCI	43	—	52.21 ± 1.06 c	—	—	—	Untargeted
Sotelo‐Orozco et al. (2020)	31	193	3.7 ± 0.76	2.0–5.0	3.59 ± 0.81	2.0–5.0	Untargeted
Montal et al. (2021)	103	71	—	31.0–56.2 b	—	49.4–57.1 b	Targeted
Patkee et al. (2021)	10	22	39.15 a	37.57–45.57 d	42.86 a	39.57‐43.86 d	Targeted
Ex vivo	14	30	—	10.0–19.0 e	—	10.0–20.0 e	Targeted
Meguid et al. (2022)	25	—	15.22 ± 1.99	—	—	—	Targeted
DS overweight/obese	25	—	16.26 ± 1.76	—	—	—	Targeted
Cai et al. (2023)	17	23	22.35 ± 11.78	4.0–45.0	24.05 ± 10.74	7.0–51.0	Untargeted
El Feil et al. (2023)	40	40	10.85 ± 2.72	6.0–15.0	10.7 ± 2.64	5.0–17.0	Targeted
Galbraith et al. (2023)	316	103	23.0 a	15.0–32.0 b	28.0 a	15.0–39.0 b	Targeted
Hetman et al. (2023)	42	20	14.17 ± 6.7	—	15.92 ± 8.58	—	Untargeted
Donovan et al. (2024)	304	96	24.05 ± 12.01	1.6–57.6	28.09 ± 14.99	4.2–61.3	Targeted
Gillenwater et al. (2024)	244	46	22.8 ± 11.8	—	22.4 ± 15.5	—	Targeted
Koenig and Bhattacharyya (2024)	17	17	44.9 ± 7.2	29.0–58.0	44.9 ± 7.1	29.0–58.0	Targeted
Santoro et al. (2025)	34	27	18.9 a	15.8–20.9 b	16.9 a	14.9–17.2 b	Untargeted

Abbreviations: AD, Alzheimer disease; CT, control; DS, Down syndrome; STD, standard deviation.

^a Age provided as median.

^b Age range provided as interquartile range.

^c Age provided as mean ± standard error of mean.

^d Age determined as postmenstrual age in weeks.

^e Age determined as postconceptional weeks.

3.2. Analytical Approaches and Methodological Heterogeneity

To further characterise the included studies, the analytical approaches and biological matrices employed were examined. The predominant analytical strategy was targeted metabolomics, used in 22 studies. Twelve studies opted for an untargeted approach, and one employed both methods. Details of the untargeted studies are provided in Table 3, while Table 4 summarises those using a targeted approach.

TABLE 3.

Description of untargeted metabolomics studies on DS population.

First author (year)	Sample type	Technique employed	Functional analysis	Data availability	Main findings
Tansley et al. (2012)	Plasma, serum	GC–MS	No	Not available	No alterations in sterol lipid metabolism were found; however, brassicasterol was significantly reduced in DS group.
Culp‐Hill et al. (2017)	RBC	UHPLC–MS	No	Supplementary xlsx file	Deregulation of RBC metabolism in DS, which may be related to comorbidities associated with DS, as well as metabolic signatures of ageing.
Caracausi et al. (2018)	Plasma, urine	1H‐NMR	No	Supplementary xlsx file	Alterations in metabolic pathways, possibly related to altered energy metabolism.
Gross et al. (2019)	Plasma	UPLC‐QTOF‐MS, GC–MS	Yes	Not available	Multiple metabolic pathways are dysregulated in the group DS‐AD compared to the DS‐nad.
Powers et al. (2019)	Plasma, CSF	UHPLC‐HRMS	Yes	Deposited	Alterations in the kynurenine pathway, suggesting a possible mechanism of immunosuppression and neurotoxicity in DS.
Orozco et al. (2019)	Plasma	1H‐NMR	No	On request	DS group presented alterations in one‐carbon metabolism, energy‐related TCA and methylation metabolism pathways, relation to mitochondrial dysfunction.
Antonaros et al. (2020)	Plasma	1H‐NMR	No	Supplementary xlsx file	There is a metabolic imbalance able to discriminate between DS and CT groups, with no apparent association with the degree of ID.
Mapstone et al. (2020)	Plasma	UPLC‐QTOF‐MS	Yes	Not available	Metabolic alterations in the context of DS may be involved in clinical and cognitive decline related to AD.
Sotelo‐Orozco et al. (2020)	Plasma	1H‐NMR	No	Supplementary xlsx file	Follow up, correlations suggest an association between altered metabolites and poorer cognitive and adaptive subscales, as well as poorer scores on a number of subscales.
Cai et al. (2023)	Faecal	GC‐TOF‐MS	Yes	Supplementary docx file	Alterations in metabolic pathways of amino acids, ketone bodies, glyoxylate and dicarboxylate and citrate cycle.
Hetman et al. (2023)	Serum	1H‐NMR	No	Not available	Metabolomic analysis suggests an imbalance that may predispose DS people to vascular complications.

Abbreviations: AD, Alzheimer disease; CT, control; DS, Down syndrome; RBC, red blood cells; TCA, tricarboxylic acid.

TABLE 4.

Description of targeted metabolomics studies on DS population.

First author (year)	Sample type	Technique employed	Functional analysis	Data availability	Main findings
In vivo metabolomics
Murata et al. (1993)	Brain	1H‐MRS	No	Not available	DS patients in the fifth decade showcase metabolic abnormalities such as degradation and rapid synthesis of brain cell membrane prior to neuronal loss and degradation.
Berry et al. (1999)	Brain	1H‐MRS	No	Not available	Elevated myo‐inositol levels in DS group.
Huang et al. (1999)	Brain	1H‐MRS	No	Not available	DS groups exhibited higher levels of myo‐inositol, possibly due to an effect of extra gene dose, which are related to predementia phase in symptomatic Alzheimer's disease.
Śmigielska‐Kuzia and Sobaniec (2007)	Brain	1H‐MRS	No	Not available	Abnormal metabolism of stimulatory amino acids related to developmental disorders and brain ageing in children with DS.
Śmigielska‐Kuzia et al. (2010)	Brain	1H‐MRS	No	Not available	Abnormal metabolism of neurotransmitters in central nervous system of children with DS.
Lamar et al. (2011)	Brain	1H‐MRS	No	Not available	DS group with dementia exhibited higher myo‐inositol levels, even when compared to AD, suggesting myo‐inositol may modify risk for dementia in the DS population.
Phillips et al. (2013)	Muscle	31P‐MRS	No	Not available	The phosphocreatine recovery rate constant lower in DS group, suggesting impaired muscle mitochondrial function.
Tan et al. (2014)	Brain	1H‐MRS	No	Not available	DS group did not have significant differences in hippocampal Glx concentrations compared to non‐DS group.
Lin et al. (2016)	Brain	1H‐MRS	No	Not available	Possible higher glial inflammatory component in DS. NAA and NAA/myo‐inositol ratio are sensitive to functional status of adults with DS.
Montal et al. (2021)	Brain	1H‐MRS	No	Not available	Higher myo‐inositol levels in DS group. Levels of myo‐inositol increased with disease progression, while NAA levels decreased.
Patkee et al. (2021)	Brain	MRS	No	Not available	Higher myo‐inositol levels in brains of DS groups, which could potentially be linked to different cognitive outcomes.
El Feil et al. (2023)	Brain	1H‐MRS	No	Not available	Significant reduction of various brain metabolites in DS group compared to control with a significant correlation between IQ and metabolic ratios in brains of children with DS.
Koenig and Bhattacharyya (2024)	Brain	1H‐MRS	No	On request	Metabolite data were not related to measures associated with dementia status in DS.
Biofluids‐based metabolomics
Shetty et al. (1995)	Plasma, CSF	GC–MS	No	Not available	Myo‐inositol concentration ratio of CSF/plasma were significantly elevated in the DS group but not correlated with dementia.
Biselli et al. (2012)	Plasma	LC–MS/MS	No	Not available	Polymorphisms present in DS group may impact folate metabolism.
Obeid et al. (2012)	Plasma, serum	UPLC‐MS/MS, GC–MS, LC–MS/MS	No	Not available	Enhanced turnover of methionine in DS compared with controls.
Mendes et al. (2013)	Plasma	LC–MS/MS	No	Not available	DHFR DD/SHMT and TT DHFR II/SHMT TT genotypes are associated with concentrations of folate and MMA in DS group.
Signorini et al. (2018)	Plasma	GC–MS/MS	No	Not available	4‐F4t‐Neuroprostane and 10‐F4t‐Neuroprostane are related with neurological conditions and may be useful for identifying pathogenic pathways in brain diseases.
Gross et al. (2019)	Plasma	LC–MS/MS	Yes	Not available	Multiple metabolic pathways are dysregulated in the DS with AD group compared to DS without AD.
Patkee et al. (2021)	Brain	GC–MS, LC–MS	No	Not available	Higher myo‐inositol levels in brains of DS groups, which could potentially be linked to different cognitive outcomes.
Meguid et al. (2022)	Blood, plasma	UPLC‐MS/MS	No	On request	CBS enzyme alongside cysteine may modulate DS obesity through increased hydrogen sulphide production.
Galbraith et al. (2023)	Plasma	UHPLC‐HRMS	Yes	Deposited	Interferon hyperactivity is accompanied by metabolic changes, where dysregulated tryptophan catabolism is prominently affected.
Donovan et al. (2024)	Plasma	UHPLC–MS	Yes	Deposited	DS individuals exhibit a depletion of circulating amino acids.
Gillenwater et al. (2024)	Plasma	UHPLC‐HRMS	Yes	Deposited	Cytokine–metabolite relationships differ between euploid and DS individuals, with clustering analysis revealing DS subgroups with varying frequencies of co‐occurring conditions.

Abbreviations: AD, Alzheimer disease; CBS, cystathionine β‐synthase; CSF, cerebrospinal fluid; DS, Down syndrome; Glx, glutamate‐glutamine complex; MMA, Methylmalonic acid; NAA, N‐acetylaspartate.

Among untargeted metabolomics studies, seven used MS analysis, while five employed ¹H‐NMR. The most common sample type was plasma, featured in nine articles, followed by serum and cerebrospinal fluid (CSF), each used in two articles. Red blood cells, urine and faecal samples appeared in one article each.

Targeted metabolomics studies were divided into two groups: (a) in vivo analyses and (b) biofluid‐based analyses. Nearly all in vivo studies used ¹H‐MRS to evaluate metabolites, with only one study employing ³¹P‐MRS; aside from a single study analysing muscle, all focused on brain tissue. Regarding biofluid‐based analyses, all employed MS techniques, including tandem MS and high‐resolution MS. Plasma was the most analysed sample type (n = 9). Meanwhile, serum, CSF, blood and brain tissue were each analysed in one study. This diversity in analytical strategies and sample types reflects both the breadth of metabolomic approaches and a source of variability across studies.

Considerable methodological heterogeneity was observed across studies, particularly in quality control procedures, normalisation strategies, metabolite identification confidence levels and pre‐analytical conditions such as cohort fasting status (Table S1). While many studies reported the use of internal standards and false discovery rate corrections, others lacked detailed reporting of key analytical parameters. This variability has important implications for cross‐study comparability and interpretation of metabolomic findings. As a result, some apparent inconsistencies across studies may reflect methodological differences rather than true biological divergence.

3.3. Overview of Metabolomic Alterations and Evidence Consistency

To facilitate interpretation of this heterogeneous evidence base, metabolomic findings were synthesised according to their level of consistency across studies. Within the body of evidence identified in this scoping review, findings can be broadly categorised into three levels of consistency: (i) findings consistently replicated across multiple independent cohorts, (ii) partially convergent findings showing variability in direction or context and (iii) isolated observations reported in single studies that require further validation. Collectively, metabolomic findings across studies reveal recurrent patterns of alteration affecting crucial biological domains, including polyol metabolism, energy metabolism, amino acid metabolism, lipid metabolism, neurotransmitter pathways and immune‐related processes (Table S2). These alterations were identified across multiple tissues and biofluids, including brain, plasma, urine and CSF, and were reported in cohorts spanning from foetal stages to adulthood.

3.4. Patterns of Metabolic Alterations Across Studies

Based on this framework, specific patterns of metabolic alterations were further examined. Among the consistently replicated findings across independent cohorts, only seven studies extended their statistical analyses to functional or pathway analyses, and relatively consistent alterations were observed. The polyol metabolism showed a predominantly increased profile across all age groups, supported by repeated observations of elevated myo‐inositol. Similarly, pathways related to immune activation and inflammation, such as the tryptophan–kynurenine axis, were consistently reported as increased across studies (Table S2 and Data S1). Neurotransmitter‐related metabolites showed decreased levels, particularly in adult populations and in Alzheimer‐type dementia. Amino acid metabolism also showed a tendency toward reduced metabolite levels across multiple cohorts, though with some variability by amino acid and study population.

In contrast, partially convergent findings were observed in several metabolic domains, showing substantial heterogeneity across studies. One‐carbon metabolism and methylation‐related pathways yielded inconsistent findings, with both increases and decreases reported across cohorts. Lipid metabolism was also widely dysregulated, with marked variability in the direction and magnitude of changes (Table S2). Alterations in energy metabolism were similarly heterogeneous; however, a recurrent pattern of dysregulation was observed. Several studies reported increased levels of tricarboxylic acid cycle (TCA cycle) intermediates, and reductions in markers of muscle energy metabolism were also noted, suggesting impaired bioenergetic efficiency despite variability in specific metabolite profiles (Table S2 and Data S1).

3.5. Directionality and Context‐Dependent Metabolomic Patterns

A more detailed assessment of the directionality of metabolite changes across studies (Table 5) further supports the observed patterns. Increased metabolites were more frequently reported in pathways related to immune activation, polyol metabolism and specific energy metabolism intermediates. In contrast, decreased metabolites were more commonly observed in neurotransmitter‐related pathways and amino acid metabolism.

TABLE 5.

Breakdown of the directionality of change of the metabolites analysed per study.

Study	Number of metabolites analysed	Number of significant metabolites	Increased metabolites	Decreased metabolites
Murata et al. 1993	3	2	Cho/Cr ratio	NAA/Cho ratio
Shetty et al. 1995	8	4	Myo‐inositol (CSF), mannitol, ribitol, arabitol (plasma)	Nonsignificant
Huang et al. 1999	4	4	Myo‐inositol, Cho‐containing compounds, total Cr	NAA
Berry et al. 1999	4	2	Myo‐inositol	Cho‐containing compounds
Śmigielska‐Kuzia and Sobaniec 2007	5	2	Nonsignificant	NAA/Cr, Glx/Cr
Śmigielska‐Kuzia et al. 2010	5	4	Nonsignificant	NAA/Cr, Cho/Cr, Myo‐inositol/Cr, GABA/Cr
Lamar et al. 2011	4	3	Myo‐inositol (compared to controls, in DS with dementia group)	NAA (specifically in DS with dementia compared to DS without)
Obeid et al. 2012	11	8	Cystathionine, cysteine, betaine, Cho, dimethylglycine, SAH, SAM	Total Hcy, SAM/SAH ratio
Biselli et al. 2012	2	—	Hcy (specifically in carriers of the MTR 2756 AG/GG genotype)	Hcy (specifically in carriers of MTHFR 677 TT, TC2 776 GG and BHMT 742 AA genotypes)
Tansley et al. 2012	15	4	Cholestanol, cholesterol (in < 45 years)	Brassicasterol, sitosterol, 27‐OH cholesterol (in < 45 years), campesterol (in > 45 years)
Mendes et al. 2013	2	—	Folate, methylmalonic acid (changes depended on specific DHFR and SHMT combined genotypes)	Nonsignificant
Phillips et al. 2013	4	—	Nonsignificant	Nonsignificant
Tan et al. 2014	1	—	Nonsignificant	Nonsignificant
Lin et al. 2016	3	3	Myo‐inositol	NAA, Glx (specifically in DS with Alzheimer's disease)
Culp‐Hill et al. 2017	225	Not explicitly stated	Lactate, amino acids (except methionine), hypoxanthine, glutamine, glutamate, 5‐oxoproline, glutathione (GSH/GSSG), citrate, succinate, fumarate, malate, alpha‐ketoglutarate, bile acids, acyl‐conjugated carnitines	Methionine, arachidic acid, behenic acid, elaidic acid, carbohydrates
Caracausi et al. 2018	33 (plasma) 30 (urine)	11 (plasma) 4 (urine)	Acetate, acetoacetate, acetone, creatine, formate, glutamine, glycerol, pyruvate, succinate (plasma); phenylacetylglycine, TMAO, tyrosine (urine)	Lysine, tyrosine (plasma); glycine (urine)
Signorini et al. 2018	2	1	10‐F4t‐neuroprostane	Nonsignificant
Orozco et al. 2019	49	7	Cho, N‐N‐dimethylglycine, dimethyl sulfone, carnitine, O‐acetylcarnitine, Cr, 2‐oxoglutarate	Nonsignificant
Gross et al. 2019	~5000 (untargeted) 4 (targeted)	5 (marked as relevant by authors)	Lactic acid, pyruvic acid, methyladipic acid (in DS with Alzheimer's disease)	Uridine (in DS with Alzheimer's disease)
Powers et al. 2019	91	29	Kynurenine, quinolinic acid, succinate, 2‐hydroxyglutarate, dehydroascorbate, mannitol, acyl‐C5:1, D‐ribose, 5‐hydroxyindoleacetate, L‐cysteine	L‐serine, L‐tyrosine, L‐homocysteine, indole, indole‐3‐acetaldehyde, (5‐L‐glutamyl)‐L‐glutamine, 3‐methyleneoxindole, sphingosine 1‐phosphate
Sotelo‐Orozco et al. 2020	49	8	Arginine, acetoacetate (correlated with better psychometric scores), lactate, pyruvate, 3‐hydroxybutyrate, acetoacetate, cis‐aconitate, fumarate, ornithine (correlated with poorer psychometric scores)	Nonsignificant
Mapstone et al. 2020	8805	180	Not explicitly reported
Antonaros et al. 2020	28	7	Acetate, pyruvate, Cr, formate, succinate, unk3	Unkown 1; tyrosine, histidine, threonine (before FDR correction)
Montal et al. 2021	2	2	Myo‐inositol	Total NAA, on symptomatic stages of disease
Patkee et al. 2021	5	3	Myo‐inositol, C, glycine	Nonsignificant
Meguid et al. 2022	4	—	Cysteine (exceeded normal ranges in both groups)	Nonsignificant
Cai et al. 2023	Not explicitly stated	35	Tyramine, methionine sulfoxide 2, 3‐hydroxybutyric acid, oxamic acid, ocalacetic acid, farnesal 1, cycloserine, glutaraldehyde 2, phenylacetic acid, phenylalanine 1, isoleucine, norvaline, indole‐3‐acetamide 2, benzyl alcohol, 4‐methyl‐5‐thiazolethanol, pipecolinic acid, beta‐alanine 2, valine, alanine 1, glutaric acid, serine 1	N‐acetyl‐D‐galactosamine 1, 4‐aminobutyric acid 1, cellobiose 2, conduritol b epoxide 2, alpha‐tocopherol, ribonic acid gamma‐lactone, 3‐cyanoalanine, asparagine 3, elaidic acid, alpha‐aminoadipic acid, arachidic acid, 1‐monopalmitin, DL‐dihydrosphingosine 1, behenic acid
El Feil et al. 2023	3	3	Nonsignificant	NAA, myo‐inositol, Cho (ratios to Cr)
Galbraith et al. 2023	174	119	Kynurenine, 5‐hydroxyindoleacetate, quinolinic acid (54 total)	Not explicitly reported (65 total)
Hetman et al. 2023	30	8	Lysine, glutamine	Acetate, proline, formate, xantine, creatinine, pyroglutamate
Koenig and Bhattacharyya 2024	6	—	Nonsignificant	Nonsignificant
Gillenwater et al. 2024	174	Not explicitly stated	Hexadecanoic acid (IMS1), acetylcholine (IMS2), IFN‐γ, SAA, IL‐29, kynurenine, 5‐hydroxyindoleacetate (IMS3), alanine (IMS4)	Not explicitly reported
Donovan et al. 2024	Not explicitly stated	9	Not explicitly reported	Asparagine, serine, proline, alanine, threonine, histidine, L‐citrulline, gamma‐L‐glutamyl‐D‐alanine (mostly prominent in MS3)
Santoro et al. 2025	Not explicitly stated	Not explicitly stated	Thymidine, butanoyl‐l‐carnitine, nicotinamide, tetradecenoic acid, heptanoic acid, octanoic acid, L‐proline, acetylcholine	4‐pyridoxate, glycerol‐3‐phosphate, glutamate, gamma‐glutamyl‐gamma‐aminobutyrate, guanosine

Abbreviations: BHMT, Betaine‐Homocysteine S‐Methyltransferase; Cho, Choline; Cr, Creatine; CSF, Cerebrospinal fluid; DS, Down syndrome; GABA, γ‐aminobutyric acid; Glx, Glutamate‐glutamine complex; GSH, Reduced glutathione; GSSG, oxidised glutathione; IFN‐γ, Interferon‐γ; IL‐29, Interleukin‐29; IMS1, Immunometabolic subgroup 1; IMS2, Immunometabolic subgroup 2; IMS3, Immunometabolic subgroup 3; IMS4, Immunometabolic subgroup 4; MS3, Molecular subtype 3; MTHFR, Methylenetetrahydrofolate reductase; MTR, methionine synthase; NAA, N‐acetylaspartate; SAA, Serum amyloid A; SAH, S‐adenosylhomocysteine; SAM, S‐adenosylmethionine; TMAO, Trimethylamine‐N‐oxide.

To further clarify the influence of clinical context, metabolomic findings were examined according to age group and phenotype (Table S2 and Data S1). Across the available evidence, patterns were not uniformly distributed, but instead reflected differences between paediatric populations, adult individuals with DS and those with dementia‐associated phenotypes. In paediatric cohorts, including foetal, neonatal and childhood populations, metabolomic alterations were most frequently reported in pathways related to amino acid metabolism, immune activation and early energy metabolism shifts. However, these findings showed greater variability across studies, particularly in lipid and one‐carbon metabolism, where directionality was inconsistent.

In adult populations without dementia, metabolomic profiles reflected a combination of persistent alterations observed earlier in life and emerging disruptions in energy metabolism and immune‐related pathways. These included continued dysregulation of amino acid and lipid metabolism, alongside evidence of altered bioenergetic pathways, although heterogeneity remained substantial across studies. In contrast, studies involving adults with Alzheimer‐type dementia demonstrated more consistent neurometabolic alterations. These included increased myo‐inositol and decreased N‐acetylaspartate, reflecting changes in polyol metabolism and neuronal integrity. Neurotransmitter‐related metabolites also showed a more pronounced decrease in these populations, supporting the notion that certain metabolomic alterations may become more evident or stabilise in later disease stages.

Across all age groups, some metabolic domains, such as polyol metabolism and immune‐related pathways, showed relatively consistent patterns, whereas others, including lipid metabolism and one‐carbon pathways, remained highly variable regardless of age or phenotype. Distinct metabolomic profiles were also reported in specific subgroups, including individuals with DS regression disorder (DSRD) and those characterised by immunometabolic phenotypes, further supporting the context‐dependent nature of metabolomic alterations in DS. These findings indicate that metabolomic alterations in DS follow age‐ and phenotype‐dependent trajectories, with certain pathways showing consistent patterns across the lifespan, while others remain heterogeneous and influenced by clinical context. Overall, the current evidence base is marked by partial convergence and substantial heterogeneity, underscoring the need for cautious interpretation and prioritisation of findings supported by independent replication.

4. Discussion

4.1. Principal Findings and Evidence Integration

This scoping review provides an integrated synthesis of metabolomic findings in DS, highlighting consistently replicated patterns, partially convergent evidence and isolated observations across studies. For that, 34 studies characterising metabolomic features in individuals with DS were identified. The evidence suggests that while certain metabolic alterations appear consistently across independent cohorts, a substantial proportion of findings remain context‐dependent and are influenced by methodological and clinical variability. Furthermore, these patterns were not uniform across the lifespan but instead reflected age‐ and phenotype‐dependent differences, particularly between paediatric, adult and dementia‐associated populations. These findings provide a structured basis to distinguish between more robust signals and those requiring further validation.

4.2. Biological Interpretation of Consistent and Heterogeneous Findings

The interpretation of metabolomic alterations in DS must consider both consistently replicated findings and those exhibiting partial convergence across studies. Collectively, disruptions in energy, polyol, one‐carbon, amino acid and lipid pathways, as well as alterations in immune, inflammatory and neurotransmitter functions, were identified (Figure 2 and Tables 5 and S2). Most studies have examined plasma or brain tissue, with only a few examining other biofluids or tissues. While these matrices are undeniably informative, expanding to different sample types could reveal additional insights into the development of health conditions. In this context, understanding how metabolic alterations relate to the broader clinical phenotype of DS is essential. For example, faecal metabolomics can uncover diet–microbiota–host interactions and has been used to study inflammatory bowel diseases (Karu et al. 2018; Vich Vila et al. 2024). Saliva and urine are practical alternatives because they are non‐invasive to collect and have potential for clinical scalability (Zhang et al. 2012; Zhao et al. 2025). Interestingly, urine can track time‐dependent changes in metabolites, making it a promising tool for real‐time health monitoring (Li 2015; Miller et al. 2019).

FIGURE 2.

Representation of altered metabolic pathways in people with DS. Coloured paths represent possible alterations caused by metabolite imbalance. Figure adapted from KEGG Pathway database (Kanehisa et al. 2025).

Since the discovery of the genetic basis of DS, considerable progress has been made in understanding the molecular mechanisms behind its related comorbidities. Recent population‐based evidence indicates that these comorbidities in DS are not randomly distributed but form distinct clusters, such as autoimmune, cardiovascular, neurodegenerative and mental health conditions, and follow age‐dependent trajectories (Baksh et al. 2023). Such patterns suggest underlying biological and potentially metabolic particularities, further supporting the idea that DS possesses unique metabolomic signatures.

Within this broader framework, several biological domains have been more extensively investigated. A long‐standing focus of study has been on the brain's metabolic phenotype in DS, especially its connection to cognitive decline, dementia and intellectual disability. The most consistently replicated alterations include increased myo‐inositol and other metabolites related to polyol metabolism. Disruptions in polyol metabolism may cause membrane instability and gliosis, while increased myo‐inositol levels have been linked to cognitive decline and AD, both highly common in DS (Shetty et al. 1995; Ali et al. 2022).

Beyond polyol metabolism, other pathways show a more complex and heterogeneous profile. For instance, although alterations in energy metabolism also display heterogeneity, a recurrent theme of mitochondrial dysfunction is supported by shifts in TCA cycle intermediates and muscle energy markers, aligning with metabolomic signatures of accelerated ageing, a feature highly associated with the DS population and their increased risk of developing early‐onset AD (Lott 1982; Zigman 2013; Gensous et al. 2019). The bioenergetic alterations highlighted by the included metabolomics studies have also been seen to occur on gene expression patterns by a recent meta‐analysis, where genes related to oxidative phosphorylation were downregulated, in addition to the dysregulation of others involved in metabolic routes such as the TCA cycle, glycolysis, gluconeogenesis and the pentose phosphate pathway, indicating a possible impairment of mitochondrial activity on individuals with DS (Pecze and Szabo 2021).

Oxidative stress, associated with early ageing, is also strongly linked to chronic inflammation and immune dysregulation, two other hallmarks of DS (Slivšek et al. 2025). Immune dysfunction is correlated with high rates of conditions such as primary immunodeficiencies, Hashimoto's thyroiditis, type 1 diabetes mellitus and coeliac disease (Pellegrini et al. 2012; Ferrari and Stagi 2021; Baksh et al. 2023). A consistent increase of compounds related to immune dysregulation was found primarily in pooled or adult cohorts, where kynurenine pathway hyperactivity and tryptophan catabolism dysregulation have been proposed as key pathways associated with interferonopathy in trisomy 21 (Powers et al. 2019; Galbraith et al. 2023). Interestingly, four of the six interferon receptors are encoded in HSA21 and have been shown to be upregulated in cells with trisomy 21 alongside proteins involved in immune control, contributing to the previously mentioned interferonopathy and chronic immune dysregulation (Sullivan et al. 2016, 2017; Illouz et al. 2021). While recurrent alterations in pathways such as polyol metabolism and immune‐related processes support their potential biological relevance, other pathways display substantial variability, limiting the strength of generalisable conclusions. Importantly, these associations should be interpreted with caution, as most evidence remains cross‐sectional and does not allow definitive mechanistic conclusions to be drawn.

In addition to metabolic alterations, epigenetic mechanisms may further contribute to the observed phenotype. Epigenetic changes, including distinct DNA methylation patterns, have also been seen in DS and are strongly linked to cognitive decline and early development of ad (Gensous et al. 2019; Yu et al. 2020). Alterations found in one‐carbon, folate and transsulfuration pathways, although inconsistent, may be associated with hypermethylation events documented by epigenome‐wide association studies and the overexpression of methyltransferases (Yu et al. 2020; Muskens et al. 2021; Vione et al. 2025).

From a broader systems perspective, these findings can be contextualised within the genetic architecture of DS. Traditionally, phenotypic variability in DS has been linked to gene‐dosage effects. Accordingly, HSA21, the second human chromosome to be fully sequenced, has been extensively studied because of its key role in DS development (Hattori et al. 2000; Antonarakis 2017). HSA21 encodes over 500 genes, and although it is not particularly rich in protein‐coding sequences, it has one of the highest densities of long noncoding RNAs among human chromosomes (Hattori et al. 2000; Letourneau and Antonarakis 2012; Antonarakis 2017; Moyer et al. 2021). Several HSA21 genes are implicated in inflammation, immunity and epigenetic regulation, and their overexpression is believed to contribute to DS‐related features. This hypothesis has led to the creation of multiple DS animal models that exhibit various aspects of the syndrome's associated pathologies (Antonarakis 2017; Yu et al. 2020; Ferrari and Stagi 2021; Moyer et al. 2021).

However, the traditional gene‐phenotype paradigm is overly simplistic. The relationship between genotype and phenotype should not be viewed as a direct cause‐and‐effect link between a single gene and a specific trait, as multiple genetic and environmental factors influence phenotypic expression (Orgogozo et al. 2015; Fisch 2017; De Vienne 2022). In DS, phenotypic diversity remains despite a consistent gene‐dosage imbalance, emphasising the role of genetic variation and gene–environment interactions. Therefore, research prioritising the phenotypic differences among individuals with DS rather than relying solely on gene‐based assumptions may better capture clinically relevant insights (Orgogozo et al. 2015; Antonarakis 2017). Together, these observations reinforce the complexity of genotype–phenotype relationships in DS and highlight the need for integrative approaches.

Within this context, metabolomics offers a unique integrative perspective. Notably, while transcriptomic and proteomic data provide insights into potential biological events, metabolomics reflects the cellular environment, capturing molecular processes that have already occurred while integrating inputs from diet, microbiota and environmental exposures, which can significantly influence health outcomes (Schmidt 2004). Metabolomic profiling at a population scale, together with machine and deep learning–based analysis tools, has been reported to be useful for identifying high‐risk groups across a wide range of high‐morbidity conditions, demonstrating the potential of metabolomics applications for population‐level risk assessment (Buergel et al. 2022; Nightingale Health Biobank Collaborative Group et al. 2024). Beyond static profiling, metabolic flux approaches, widely used in metabolic engineering, could help clarify pathway directionality and control points in DS by resolving rates across interconnected networks (Nielsen 2003; Wittmann and Portais 2013).

4.3. Methodological Considerations and Sources of Variability

An important consideration in interpreting these findings is the substantial methodological variability across studies. Differences in analytical platforms, metabolite identification confidence, sample handling and data processing strategies may significantly influence reported results, and in some cases, apparent inconsistencies may reflect methodological factors rather than true biological divergence. Consequently, despite the emerging patterns, the strength of evidence supporting individual findings remains limited and varies considerably across studies.

Also, although extensive, metabolomics profiling of individuals with DS, as evidenced in this scoping review, has primarily been performed using a cross‐sectional design, leaving the resulting data in an associative state. Thus, mechanistic inferences remain hypothetical as causality cannot be clearly established by the current reported data. In this regard, longitudinal metabolomics research may help uncover associations between metabolites of interest, viewed as potential disease predictors, and observed outcomes.

4.4. Clinical Implications and Future Directions

The potential clinical implications of these findings should be interpreted cautiously. Integration with other omics layers, such as transcriptomics and proteomics, may enhance biological interpretation; however, such integrative approaches remain limited and require validation within well‐characterised cohorts. Notably, alterations in the tryptophan–kynurenine pathway are consistent with transcriptomic signatures of interferon pathway activation reported in trisomy 21 (Sullivan et al. 2017; Powers et al. 2019; Galbraith et al. 2023; Donovan et al. 2024), supporting the concept of a chronic interferon‐driven immunometabolic state. However, such cross‐omics concordance remains limited by the scarcity of studies integrating metabolomics with transcriptomic or proteomic data within the same cohorts. As a result, while these parallels provide biological plausibility, they should be interpreted cautiously and highlight the need for integrative, multilayered approaches to better define mechanistic pathways and identify potential biomarkers in DS.

In this regard, metabolomics, especially when used within multi‐omics and systems biology approaches, may offer potential opportunities to deepen our understanding of DS pathophysiology. By linking genetic differences, environmental factors and phenotypic traits, while incorporating longitudinal designs alongside standardised analytical pipelines, researchers may, in the future, help identify metabolic signatures relevant to risk stratification and early detection of age‐related comorbidities, potentially revealing novel biomarkers tailored to people with DS. However, current evidence remains insufficient to support clinical application. Therefore, these interpretations should be considered hypothesis‐generating, as most of the available evidence comes from cross‐sectional and correlational studies that do not allow causal inference.

5. Limitations

As a scoping review, this study did not formally assess risk‐of‐bias. Heterogeneity across matrices, life stages, analytical platforms and designs was described but not statistically reconciled. Restrictions on cross‐study comparability further arise from the methodological variability as observed in Table S1. The wide array of analytical platforms employed across studies has varying sensitivities and detection limits; consequently, a metabolite's lack of significance in one study (especially in untargeted metabolomics studies) may reflect platform‐specific threshold limitations rather than a true biological consensus. Moreover, many studies relied on putative annotations of metabolites rather than unambiguous identification utilising authentic chemical standards, increasing the risk of misidentification. Accordingly, future research should adopt standardised procedures, such as minimising short‐term storage, centrifuging before processing or controlling the number of freeze–thaw cycles (Bernini et al. 2011; Emwas et al. 2015, 2025; Kirwan et al. 2018; González‐Domínguez et al. 2020), to mitigate potential confounding effects arising from methodological inconsistencies and to ensure that findings are amenable to future quantitative integration. Consequently, differences across studies should not be interpreted solely as biological variability, as they may reflect underlying methodological inconsistencies.

Additionally, our inclusion scope (human participants profiles with MS, NMR or MRS) enhances clinical relevance but might exclude relevant data obtained through other methods or experimental models that clarify mechanisms and causality. Some of the included studies measured metabolic factors but were not explicitly designed to examine metabolic imbalances, which may limit the interpretation of their findings in the context of DS pathophysiology. Furthermore, most datasets focused on brain and plasma measurements, leaving saliva, urine, stool and peripheral tissues comparatively underexplored.

Another notable limitation involves data availability. Several studies lacked an explicit data availability statement; some offered limited supplementary files, while only a few provided accession records for raw data. This probably reflects the emerging nature of metabolomics as a field. However, we strongly encourage future metabolomics researchers to deposit raw and processed data with rich metadata in public repositories to improve reproducibility and support data integration across cohorts and platforms.

6. Conclusion

Taken together, the available evidence indicates that metabolomic alterations in DS are characterised by partially convergent yet context‐dependent patterns affecting key biological domains. By identifying and understanding how metabolites interact in biofluids, cells and tissues, scientists can begin to explain the molecular mechanisms behind disease development. While certain alterations are consistently reported across independent cohorts, others, including lipid and one‐carbon metabolism, show substantial variability in both direction and magnitude. However, the identification of these metabolic signatures suggests that metabolomics holds promise as a framework to improve the understanding of genotype–phenotype relationships in DS and could inform the development of biologically meaningful biomarkers. Therefore, future research should adopt an integrative approach to better understand the multifaceted nature of metabolic changes in this population. One aim of a scoping review is to assess whether a systematic review is feasible. While the growing body of evidence and emerging patterns of partial convergence suggest that this may be achievable, current methodological variability still poses important challenges. Future standardisation in study design and reporting will be essential to support a robust systematic review. Overall, while metabolomic research in DS reveals promising avenues for understanding disease mechanisms, the current evidence base is characterised by partial convergence and substantial heterogeneity, emphasising the need for cautious interpretation and further validation. This underscores the importance of prioritising findings supported by independent replication and standardised methodologies.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Methodological procedures retrieved from the included studies.

Table S2: Global patterns of convergence and divergence across studies.

Data S1: Supporting information.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.