A thorough review of literature was done using an extensive exploration employing keywords and subject headings, from which a list of potential biomarkers could be created. Six electronic databases were searched for articles published up to December 2023: PubMed, Ovid Medline, Embase, Scopus, Science Direct, and Cochrane. The search criteria included “Parkinson,” “Parkinson’s Disease,” “metabolites,” “metabolomics,” “metabolic,” “metabolome,” “biomarkers,” “urine-based”, “urinary” and “human” in combination. Terms were looked for using Boolean search in titles, abstracts, and MeSH keywords to provide a specific search result.

By comparing the articles to the inclusion and exclusion criteria, the eligibility of the articles was established. Articles that did not fit the requirements for inclusion were rejected. A second round of screening was carried out using the full-text content of the articles for the final shortlisting of the studies that met the inclusion criteria.

Two reviewers (NRD & DJ) searched and assessed for the eligible studies. They extracted the important data and mentioned in Table 1, based on the listed eligibility criteria. Any disagreement was resolved after a discussion with the third reviewer (SS). Each separately extracted the data in compliance with the study’s methodological characteristics, interventions, participants, outcomes, and conclusions. PRISMA 2020 flowchart outlined the complete selection process.

The case-control quality evaluation tool from the National Heart, Lung, and Blood Institute (NHLBI) was used to evaluate the study’s overall quality (National Heart et al., 2013). NHLBI developed a set of tailored quality assessment tools in 2013 to assist reviewers in focussing on concepts that are key to a study’s internal validity. The tools are specific to certain study designs and tested for potential flaws in study methods or implementation. Based on the study design (case-control study), this quality assessment tool was used. The study quality was evaluated by two authors (NRD and NP) separately and was rated as “yes,” “no,” or “not applicable.” A study is considered of “Good” quality if it meets all or most of the specified criteria, with minimal risk of bias. A study classified as “Fair” quality does not meet the specified criteria’s and suggests some risk of bias but not enough to completely undermine the results. A “Poor” quality study has significant methodological flaws that increase the risk of bias and limit the validity of its findings. A third reviewer (SS) was brought in to help address any disputes or issues.

5,377 items were found during a primary literature search. The qualifying criteria were followed in the screening of 40 articles. Seven (Luan et al., 2015a; Bai et al., 2021; Luan et al., 2015b; Chung et al., 2023; Wichit et al., 2021; Kumari et al., 2020a; Luan et al., 2015c) of these studies satisfied the requirements for inclusion in the review. A PRISMA flow chart is shown for the search parameters, study selection, and study exclusion in Figure 1.

This review was based solely on studies with urinary metabolomic biomarkers in early and mid-onset PD. Out of total 40 studies on PD biomarkers, only 7 studies were included based on selection criteria. All studies passed the NIH quality assessment evaluation and were of “fair” quality (Table 2).

Total number of PD patients in all studies who participated was 512 and total number of healthy age and gender matched control volunteers was 385. 321 were male and 191 were female patients. A summary of the type of samples and identified UMB is described in Figure 2. P-value (mann-whitney U test), Mean decrease in accuracy (MDA), Fold change (FC) value more >1.2 indicating upregulation in PD, and Area under the ROC curve (AUC) value (determined by Youden index) was kept at 0.59 to 0.95 indicating potential metabolites (Supplementary Table S1).

All studies focussed on identifying urinary biomarkers for IPD. All three studies conducted by Hemi Luan et al., and Chung SH et al. used LC-MS to identify metabolites. LC-MS is faster, highly sensitive and capable of detecting low-abundance molecules than Nuclear Magnetic Resonance. Reproducible findings about the metabolic alterations in PD in all three studies contribute to the understanding that oxidative stress, energy metabolism and neurotransmitter metabolism pathways act as a target for biomarkers. Reproducibility was also indicated by the occurrence of common UMB in these studies. The availability of similar UMB with matched Fold Change (FC) could determine a potential urinary biomarker in early (acetylphenylalanine, hydroxytryptophan, tyrosine, kynurenine, cortisol, tiglyglycine, phenylacetylglutamine, kynurenine) and mid-onset (hydroxytryptophan, tyrosine/hydroxyphenylacetic acid, phenylacetylglutamine) PD.

From total 685 genes, 169 genes were selected from the Parkinson’s disease (C0030567) dataset in DisGeNET, prioritized based on their Disease Specificity Index (DSI). Genes with a DSI closest to 1, indicating higher specificity to Parkinson’s disease, were given preference for further analysis.

The selected genes were analyzed using the STRING database, revealing a protein-protein interaction (PPI) network in which 158 genes with 158 nodes (proteins) and 1,132 edges (interactions) shows connection and the left are disconnected and hidden. The PPI network exhibited a highly significant enrichment, with a p-value of <1.0e-16 and obtained from a random set of proteins which is a distinguishable property of STRING analysis, suggesting that these proteins are more interconnected than expected by chance, pointing to their potential roles in Parkinson’s disease pathology (Figures 3, 4). The bubble plot representing disease-gene associations and gene ontology-enriched biological pathways, with significance level is shown in Supplementary Figure 6-S2.

Using the CytoHubba plugin, ten hub genes with the highest interaction scores were identified within the PPI network. These genes, marked in red, demonstrated the highest degree of connectivity, making them strong candidates for further study in Parkinson’s disease. Some of these genes are also emerging as relevant to urinary biomarkers for Parkinson’s disease.

The top hub genes and their interaction scores are listed below:

These hub genes, particularly VPS35, VPS29, and GAK, hold potential not only as molecular drivers of Parkinson’s disease (Figure 5) but also as candidates for developing non-invasive urinary biomarkers for early diagnosis and therapeutic targeting.

Alanine derivatives were found elevated in early-stage PD indicating disrupted phenylalanine and tyrosine metabolism, which may be linked to PD patients’ reduced ability to synthesize dopamine (Luan et al., 2015a; Luan et al., 2015b; Luan et al., 2015c). There is a metabolite shift in the phenylalanine metabolism in PD from tyrosine production to producing trancinnamate instead (Shebl et al., 2024). This alteration may deprive the body of synthesizing dopamine, norepinephrine, and every other phenylalanine and tyrosine metabolite. Altered acetylphenylalanine levels cause phenylalanine metabolism to be disrupted in early-stage Parkinson’s disease and can be considered as a dopamine precursor (Ramdani et al., 2015), indicating its therapeutic role in the management of PD study have shown a statistically significant (p = 0.05) increase in acetyl phenylalanine in PD patients’ urine with fold change ranging from 1.2 to 2.5 between patients and controls (Sandler et al., 1969), and supposed to be correlated with various neurological conditions (Kobayashi et al., 1984).

This study suggested that urine KYN levels distinguished patients with early-stage PD and those with healthy controls. Urine KYN levels positively correlated with Hoehn-Yahr stages and disease durations, and negatively correlated with MMSE scores of patients with PD. These results suggested that urine KYN may represent a novel biomarker for detecting early-stage PD and evaluating the progression of PD. The urine KYN measurement might assist clinicians in identifying patients with early-stage PD (Bai et al., 2021). The kynurenine pathway has been linked to oxidative stress, immune system dysregulation, and excitotoxicity, all processes involved in PD pathology. Under the induction of pro-inflammatory cytokines, Kynurenine Pathway activation disrupts the balance between neuroprotection and neurotoxicity branches, tending to produce neurotoxic products 3-Hydroxyuric acid (a strong inducer of neurotoxic free radicals) and Quinolic acid, which play an important role in the pathogenesis of PD through excitotoxicity, oxidative stress, and/or inflammatory reactions (Chen and Geng, 2023). These changes in the pathology correspond to the early stage of PD. Neuroprotective metabolite kynurenic acid (KYNA) is reduced due to this alteration in the kynurenine pathway. KYNA analogs are important therapeutic agents in preventing or delaying PD.

Tyrosine is metabolized through several pathways, one of which leads to the production of hydroxyphenyl acetic acid. Hydroxyphenyl acetic acid is a downstream metabolite of tyrosine that results from its breakdown. In the early stages of Parkinson’s disease, the loss of dopamnergic neurons in the substantia nigra leads to reduced conversion of tyrosine into dopamine. As a result, the levels of tyrosine or its metabolites may become dysregulated. Studies suggest that alterations in tyrosine hydroxylase activity, the enzyme responsible for converting tyrosine to L-DOPA, may be detectable in early PD (Bogdanov et al., 2008; Hatano et al., 2016). A reduction in enzyme activity could indicate the early neurodegeneration process. Urinary and plasma metabolomic studies have identified elevated or altered levels of tyrosine and its metabolites in individuals with PD, which may occur even in the early stages of the disease. Tyrosine metabolism generates reactive oxygen species (ROS), contributing to oxidative stress, which is a key feature in the pathogenesis of PD (Fedorova et al., 2021).

Early research on the variance in steroidogenesis metabolism revealed that people with Parkinson’s disease (PD) had noticeably greater levels of cortisol in their saliva and blood (Charlett et al., 1998; Djamshidian et al., 2011; Hartmann et al., 1997). Stress-related cortisol release may be able to pass across the blood-brain barrier and bind to glucocorticoid receptors in the central nervous system. Cortisol excretion in the urine is thought to be a sign of elevated oxidative stress, which exacerbates dopamine cell degeneration in Parkinson’s disease (PD) (Joergensen et al., 2011).

Isoleucine and leucine levels in PD patients’ urine are strongly linked with their stage of the disease. Tremors, twitching, and muscular atrophy may be caused by a deficiency in leucine and isoleucine (Mally et al., 1997; Molina et al., 1997). Succinate functions as a junction point for multiple metabolic pathways and is involved in the production and removal of reactive oxygen species (Tretter et al., 2016).

Similar metabolites were found elevated in serum and saliva samples in PD patients (Kumari et al., 2020b). Patients with Parkinson’s disease have been shown to have higher concentrations of alanine, acetate, and trimethylamine N-oxide in both their serum and saliva samples (Kumari et al., 2020b; Kumari et al., 2020c). While alanine, acetate, and trimethylamine N-oxide content were considerably higher in both serum and saliva samples in patients with Parkinson’s disease (PD), phenylalanine, citrate, glutamine, and isoleucine were shown to be differentially raised in urine and saliva samples (Kumari et al., 2020b; Babu et al., 2018).

The PPI network for the selected 158 genes also exhibited highly significant connection with the identified urinary biomarkers. Ten hub genes with the highest interaction scores were identified within the PPI network. These genes, particularly VPS35, VPS29, and GAK, are central regulators of metabolic pathways that influence PD progression. For instance, genes such as VPS35 and VPS29, which regulate protein trafficking and degradation, may influence the levels of excreted biomolecules detectable in urine. Similarly, GAK and RAB29 are implicated in mitochondrial dysfunction, which has been linked to urinary biomarker research in Parkinson’s disease. The strong correlation between these genetic hub genes and urinary biomarkers suggests a genetic basis for the observed metabolic alterations in PD which is a novel finding. The review is one of the knowledge-enhancing chapters with few limitations. This review is conducted on early and mid-stage IPD patients’ controlled studies, hence making it less applicable to advanced stages of PD. Also, this review studied only urine based biomarkers, although urine sampling and analysis is more convenient than other methods of analysis. Small sample size and the incorporation of 3 studies (Hemi Luan et al.) by the same author leading to possibility of population overlapping and biased results are few limitations of this study. Although, two of the studies (Luan et al., 2015b; Luan et al., 2015c) had same patient population but not acknowledged anywhere in text. Moreover, they used different methods for sample analysis. As identification of metabolomics and their interaction with drug compounds has been recognised as a crucial part of drug repurposing. So, various advanced models like Graph CPIs, Graph Contrasting learning, Regular Wave Graph and heterogenous information network learning model can be used in future research to address the clinical utility of identified biomarkers, and to explore these gene-metabolite interactions to develop targeted therapies, thereby improving early detection and disease management strategies for PD (Zhao et al., 2025; Zhao et al., 2024; Cui et al., 2025; Chen et al., 2023).