Analysis of Metabolites in Gout: A Systematic Review and Meta-Analysis

(1) Background: Many studies have attempted to explore potential biomarkers for the early detection of gout, but consistent and high levels of evidence are lacking. In this study, metabolomics was used to summarize the changes of metabolites in the literature and explore the potential value of metabolites in predicting the occurrence and development of gout. (2) Methods: We searched the databases including the EMBASE, the Cochrane Library, PubMed, Web of Science, VIP Date, Wanfang Data, and CNKI, and the screening was fulfilled on 30 July 2022. The records were screened according to the inclusion criteria and the risk of bias was assessed. Qualitative analysis was performed for all metabolites, and meta-analysis was performed for metabolite concentrations using random effects to calculate the Std mean difference and 95% confidence interval. (3) Results: A total of 2738 records were identified, 33 studies with 3422 participants were included, and 701 metabolites were identified. The qualitative analysis results showed that compared with the healthy control group, the concentration of 56 metabolites increased, and 22 metabolites decreased. The results of the meta-analysis indicated that 17 metabolites were statistically significant. (4) Conclusions: Metabolites are associated with gout. Some specific metabolites such as uric acid, hypoxanthine, xanthine, KYNA, guanosine, adenosine, creatinine, LB4, and DL-2-Aminoadipic acid have been highlighted in the development of gout.


Introduction
Gout is a metabolic immune disease characterized by disorders of purine and uric acid (UA) metabolism. It is caused by the super-saturation of UA, which crystallizes in joints, leading to peripheral joint synovitis and severe pain [1,2]. With the increasing number of acute gout attacks, the patient gradually progresses to joint damage, deformity, chronic use-related pain, subcutaneous tophi deposits, and renal damage. Additionally, gout is often associated with metabolic syndrome [3], cerebrovascular diseases [4], cardiovascular diseases [5,6], and renal damage [7]. The prevalence of gout ranges from approximately 1% to 6.8%, with an incidence rate of 0.58 to 2.89 per 1000 persons per year [8]. Moreover, it has doubled in less than 30 years [9]. Studies have shown that gout patients have high healthcare costs due to frequent emergency room visits, hospitalizations, reduced productivity, and disability [10][11][12], which range from USD 172 to USD 6179 [13].
The clinical signs of gout are usually severe pain, edema, redness, and inflammation of the joints, which occur suddenly during the night and then gradually diminish and disappear over a period of days or weeks. Currently, the clinical diagnosis of gout is based on serum uric acid detection, joint synovial fluid examination, and imaging examination [14]. As gout is often acute and easily confused with other forms of arthritis, these diagnostic methods do not seem to be adequately timely and accurate [15], and thus cause patients

Study Selection
All identified records were downloaded to Endnote X9, and then the duplicates were removed by computer and manual deletion. Two independent researchers (LYY and TJL) screened the studies by title and abstract, respectively, and then the studies meeting our criteria were found in the full text for further screening. Any disagreements were discussed with the third researcher (TJL) until the team reached a consensus. A senior researcher (ZHY) gave the guidance and supervised.

Risk of Bias Assessment
Two researchers (LYY and HX) assessed the risk of bias in the included records independently by the Newcastle-Ottawa Scale (NOS), a tool for assessing the quality of observational studies [43]. It evaluates research through a "star system" based on selectivity, comparability, and exposure.

Data Collection
The extracted research data included the following items: first author's name, publication year, country of origin, study design type, language, age, sample size, diagnostic criteria of patients, statistical methods, detection techniques, metabolomics techniques, and trends and concentrations of metabolites in gout and healthy controls. Data were obtained from the article and the Supplementary Material. When the data is presented graphically only and the original author cannot be contacted for details of the data, we use Web Plot Digitizer (Web Plot Digitizer, V.4.2, San Francisco, California: Ankit Rohatgi, 2019) to extract data from the graph of the article. For those articles with validation cohorts, data from the validation cohorts were also included.
We use PubChem (https://pubchem.ncbi.nlm.nih.gov/ (accessed on 20 January 2023)) to search the metabolites which need to convert their units, and the molecular weight (g/mol) of them are shown in Supplementary Table S2. The quartile and median or interquartile range (IQR) were converted into mean and standard deviation (SD). For median and quartile, we tested the skewness first and applied a new piecewise function based on the size of the sample [36,[44][45][46][47][48]. For median and IQR, formula 1 was applied. When a study matches two or more experimental groups to one control group, the mean and standard deviation of the experimental group are combined according to formula 2. All formulas we applied were derived from the Cochrane Handbook of Systematic Reviews of Interventions and are shown in Supplementary Table S3.

Data Synthesis
A qualitative analysis was performed for changing the direction of metabolites by counting the frequency across the studies. Afterward, the concentrations of metabolites performed the meta-analysis across studies using the standardized mean difference of 46 metabolites (SMD) and confidence intervals of 95% (95% CI). Heterogeneity was deemed substantial if I 2 was greater than 60%. While in this study, a random effects model was chosen due to the heterogeneity among the records. Funnel plots and Egger's test were used to assess publication bias when feasible (10 or more studies) [49,50]. Finally, all data synthesis was performed using R software with a meta package (version 3.6.2).

Literature Search
A total of 2738 records were eventually identified; 902 duplicate records were deleted. Based on the title and abstract screening, 1733 unrelated studies were removed. In the remaining 103 studies, 71 studies were removed for reasons including not providing full text, insufficient data, and failure to meet our criteria. Ultimately, we included 33 studies [28,[33][34][35][36][37][44][45][46][47][48] and 3422 participants (2149 gout patients and 1273 healthy controls). The flow diagram for the assessment of studies identified in the systematic review is shown in Figure 1.

Literature Search
A total of 2738 records were eventually identified; 902 duplicate records were deleted. Based on the title and abstract screening, 1733 unrelated studies were removed. In the remaining 103 studies, 71 studies were removed for reasons including not providing full text, insufficient data, and failure to meet our criteria. Ultimately, we included 33 studies [28,[33][34][35][36][37][44][45][46][47][48] and 3422 participants (2149 gout patients and 1273 healthy controls). The flow diagram for the assessment of studies identified in the systematic review is shown in Figure 1.

Characteristics of Included Studies
The included records were published between 1999 and 2021 and all of them were case-control designs, among which 18 studies and 15 studies were reported in English and Chinese, respectively. Participants were from the Slovak Republic, the United States, and China. Age and sex were well matched in all studies, except that 4 studies did not report age, and 7 studies only reported age in the gout group. The studies were classified according to different sample types, including 23 blood samples, 2 urine samples, 2 stool samples, 1 saliva study, and 5 hematuria studies [33,36,47,66,71]. All the included studies measured metabolites mostly using liquid chromatography-mass spectrometry (LC-MS) or gouts chromatography-mass spectrometry (GC-MS) techniques. The specific characteristics of each study are shown in Table 1.
A total of 701 metabolites or their ratios in 33 studies were extracted. Of these, 104 metabolites occur twice or more and can be qualitatively synthesized. The 46 metabolites were distributed across 11 studies with accurate concentration values and units for metaanalysis, among which 9 studies [33,36,37,48,54,67,68,71,72] were directly obtained from the article, and 2 studies [44,46] were extracted by software from the box plots. Median and quartiles were transformed to mean and SD in 3 studies [44,46,69], and median and IQR were transformed in 3 studies [44,46,69]. Six studies [36,[44][45][46][47][48] combined the mean and SD. The characteristics of these metabolites are shown in Supplementary Table S4.

Characteristics of Included Studies
The included records were published between 1999 and 2021 and all of them were case-control designs, among which 18 studies and 15 studies were reported in English and Chinese, respectively. Participants were from the Slovak Republic, the United States, and China. Age and sex were well matched in all studies, except that 4 studies did not report age, and 7 studies only reported age in the gout group. The studies were classified according to different sample types, including 23 blood samples, 2 urine samples, 2 stool samples, 1 saliva study, and 5 hematuria studies [33,36,47,66,71]. All the included studies measured metabolites mostly using liquid chromatography-mass spectrometry (LC-MS) or gouts chromatography-mass spectrometry (GC-MS) techniques. The specific characteristics of each study are shown in Table 1.
A total of 701 metabolites or their ratios in 33 studies were extracted. Of these, 104 metabolites occur twice or more and can be qualitatively synthesized. The 46 metabolites were distributed across 11 studies with accurate concentration values and units for meta-analysis, among which 9 studies [33,36,37,48,54,67,68,71,72] were directly obtained from the article, and 2 studies [44,46] were extracted by software from the box plots. Median and quartiles were transformed to mean and SD in 3 studies [44,46,69], and median and IQR were transformed in 3 studies [44,46,69]. Six studies [36,[44][45][46][47][48] combined the mean and SD. The characteristics of these metabolites are shown in Supplementary Table S4.

Risk of Bias of Included Studies
The risk of bias in all included studies was assessed using the NOS scale. The quality of the included studies was generally high with 22 studies (66.7%) receiving over 7 stars. Five studies received 6 stars due to low exposure scores [28,48,51,53,54]. The remaining 4 studies scored less than 6 stars due to inadequate case definition [45,55,56,62]. Results on the risk of bias in the included studies and details of the evaluation of the included studies are presented in Table 1 and Supplementary Table S5, respectively.

Qualitative Synthesis
A total of 701 metabolites were identified in 33 studies, of which 567 were in blood samples, 65 were in urine samples, 49 were in fecal samples, and 20 were in saliva samples. Qualitative synthesis was performed on 104 metabolites that were reported more than twice, including 94 blood samples, 5 urine samples, and 1 fecal sample. Based on the sample type classification, 45 metabolites showed increased concentrations, 20 showed decreased concentrations, and 33 showed inconsistent concentrations in blood samples. In urine samples, 2 metabolites showed a decrease in concentration and 3 metabolites showed inconsistent concentrations. In fecal samples, only the concentrations of taurine increased.
In total, 45 metabolites had increased concentrations, 23 had decreased concentrations, and 36 showed inconsistent trends. The details of the qualitative synthesis are given in Table 2. Information about metabolites with inconsistent trends in blood (A) and urine (B) samples is shown in Figure 2.
Effect estimates were relatively consistent across studies, although substantial heterogeneity was found in some metabolite comparisons. We conducted sensitivity analyses for studies with a high risk of bias. However, the results did not change significantly from previous results. At the same time, due to the omissions in the order of individual studies, no study has greater sensitivity.      Effect estimates were relatively consistent across studies, although substantial heterogeneity was found in some metabolite comparisons. We conducted sensitivity analyses for studies with a high risk of bias. However, the results did not change significantly from previous results. At the same time, due to the omissions in the order of individual studies, no study has greater sensitivity.

Discussion
This paper summarized all studies on metabolite changes in gout patients and identified a characteristic metabolite profile for gout. The meta-analysis found 10 metabolites with significant differences and almost all of them were associated with inflammation.
Since hyperuricemia is the largest risk factor for gout, exploring the production and excretion of UA is fundamental to understanding gout. As the degradation product of purine, the main sources of UA are cell turnover, dietary intake, and de novo synthesis [73], and the primary excretion mode for UA is renal excretion [74][75][76]. Previous studies have shown that decreased renal excretion is the main cause of elevated UA, and high levels of UA are strongly associated with kidney disease [77], yet the mechanism is not clear. We aimed to investigate the role of UA and kidney in gout so that we could find reliable evidence to explain the sequence and pathological mechanism. UA induces vascular lesions characterized by endothelial dysfunction, medial wall thickening, and macrophage infiltration, reducing the blood supply to the renal tubular area and causing ischemic injury in the renal tubular area [7,78]. In addition, studies have shown that the cellular communication between renal tubules and immune cells in the tubulointerstitium plays a key role in the progression of kidney disease [79]. In this process, UA seems to accelerate the interaction between these cells, leading to persistent inflammation and progression of tubulointerstitial fibrosis.
In order to better understand the relationship between UA and the kidney, we need to understand the relationship between the two major systems of metabolism and immunity beforehand. From a survival and evolutionary perspective, the most important ability of a species is to resist starvation and mount an effective immune response to pathogens. To maximize energy efficiency, metabolism, and immunity use the same or overlapping key regulatory molecules and signaling systems [80]. In fact, functional units that control critical metabolic and immune functions in higher organisms evolved from a common ancestral structure, meaning that nutrients can act via the pathogen-sensing system to elicit metabolic or nutrient-induced inflammatory responses. In the long history of human existence, faced with a chronically unstable food supply, the human body has automatically evolved a mechanism to store excess calories. However, with the improvement of human living standards, this once beneficial bond leads to excessive accumulation of nutrients and inflammation in the absence of pathogen infection, this activation is clearly harmful, and the disorder of UA metabolism is one of them. Excessive metabolism produces more UA and activates inflammation, which is the main biological process of kidney disease [81][82][83], and meta-inflammation is the mediator to understanding the close relationship between UA and kidney disease [77,84,85]. In summary, we argue that metabolic excesses of energy and lipids lead to hyperuricemia and activate chronic metabolic inflammation. The excretion of high UA from the kidneys is the result of metabolic inflammation, and UA is both a consequence and a precipitant of renal impairment.
Confirmation of the existence of the same signaling pathway associated with UA and inflammation in gout patients is a prerequisite to proving this conjecture. Our study supports this hypothesis by showing creatinine and DL-2-aminoadipic acid were significantly different between gout and healthy controls. As an intermediate metabolite of lysine metabolism, DL-2-aminoadipic acid plays an important role in the regulation of glucose and lipid metabolism, and studies have shown that DL-2-aminoadipic acid enhances the body's metabolism and associates with damage to renal function in gout patients [46]. Therefore, the increase in DL-2-aminoadipic acid may act as a compensatory mechanism to promote the body to better carry out excess energy and fat consumption. Hypoxanthine has been shown to have the potential to cause renal inflammation in addition to its various pathophysiological consequences in humans [86,87]. As a purine metabolite, adenosine acts primarily on G protein-coupled receptors and plays an anti-inflammatory role in regulating the onset and remission of gout [88,89]. Specifically, it binds to different purinergic receptors to regulate IL-1β secretion, which is involved in the pathogenesis of gout attack [88]. Kynurenic acid (KYNA), as a branch product of the tryptophan metabolism pathway, is involved in the regulation of the cells of the immune system and a variety of immune-mediated diseases [90]. According to the newest research, chronic stress or mild inflammation can promote the production of KYNA and promote various immunomodulatory actions due to KYNA-mediated signaling pathways [91].
Having explored the meta-inflammatory basis of gout flares, we then turn to inflammation in the conventional sense caused by sodium urate crystals. In a vicious cycle of metabolic surplus and inflammation, eventually crystals precipitate and gout arthritis occurs. Sodium urate can promote the secretion of pro-inflammatory cytokines, which act on the kidneys, leading to a further decline in kidney function [66,83]. At the same time, monosodium urate crystals precipitate in the less-perfused joints and then induce an inflammatory response in the immune system [92,93]. Studies by RAE SA et al. have shown that LB4 is an important chemical mediator in acute gout attacks [94]. LB4 stimulates the production of multiple pro-inflammatory cytokines and mediators, and pharmacological data suggest that it could enhance and prolong tissue inflammation [95].
Metabolic inflammation and inflammation induced by sodium urate crystals are the basis for understanding the pathology of gout. Our study also confirmed the importance of inflammation in the pathogenesis of gout. The present study suggests that the signaling pathways used by metabolic inflammation and traditional inflammation are relatively consistent [96], starting from the root of energy metabolism and finding related biomarkers. It may be used as a new idea for early detection and prevention of gout attacks.

Limitations of the Study
Our study does have some limitations. (1) The main limitation of the study is that, while the direction of effect estimates is relatively consistent, the results of the meta-analysis do have a high degree of heterogeneity. Heterogeneity is inevitable due to different study populations, different body conditions, and different methods of metabolite detection.
(2) Although we use a new method of partition functions in converting sample sizes to means and standard deviations, the distribution of the data is still problematic. In further studies, we hope to better address this issue. (3) There was no search of the clinical trial registry database, which may have resulted in an inadequate comprehensive search.

Conclusions
Changes in metabolites in humans do shed light on the underlying mechanisms of gout development. Some specific metabolites such as UA, hypoxanthine, xanthine, KYNA, guanosine, adenosine, creatinine, LB4, DL-2-Aminoadipic acid have been highlighted in the development of gout, establishing the potential of metabolites as predictive biomarkers of gout. However, these findings require further investigation and validation in larger prospective cohort studies.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/nu15143143/s1, Table S1: Search strategy; Table S2: The molar mass of each bile acid for transforming unit; Table S3: (A) Formula 1 for converting median and interquartile range (IQR) into mean and standard deviation (SD); (B) Formula 2 for combining mean and SD;

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data reported in this review are all freely available through a database search using the included search terms.

Conflicts of Interest:
The authors declare no conflict of interest.