Arginine, Transsulfuration, and Folic Acid Pathway Metabolomics in Chronic Obstructive Pulmonary Disease: A Systematic Review and Meta-Analysis

There is an increasing interest in biomarkers of nitric oxide dysregulation and oxidative stress to guide management and identify new therapeutic targets in patients with chronic obstructive pulmonary disease (COPD). We conducted a systematic review and meta-analysis of the association between circulating metabolites within the arginine (arginine, citrulline, ornithine, asymmetric, ADMA, and symmetric, SDMA dimethylarginine), transsulfuration (methionine, homocysteine, and cysteine) and folic acid (folic acid, vitamin B6, and vitamin B12) metabolic pathways and COPD. We searched electronic databases from inception to 30 June 2023 and assessed the risk of bias and the certainty of evidence. In 21 eligible studies, compared to healthy controls, patients with stable COPD had significantly lower methionine (standardized mean difference, SMD = −0.50, 95% CI −0.95 to −0.05, p = 0.029) and folic acid (SMD = −0.37, 95% CI −0.65 to −0.09, p = 0.009), and higher homocysteine (SMD = 0.78, 95% CI 0.48 to 1.07, p < 0.001) and cysteine concentrations (SMD = 0.34, 95% CI 0.02 to 0.66, p = 0.038). Additionally, COPD was associated with significantly higher ADMA (SMD = 1.27, 95% CI 0.08 to 2.46, p = 0.037), SDMA (SMD = 3.94, 95% CI 0.79 to 7.08, p = 0.014), and ornithine concentrations (SMD = 0.67, 95% CI 0.13 to 1.22, p = 0.015). In subgroup analysis, the SMD of homocysteine was significantly associated with the biological matrix assessed and the forced expiratory volume in the first second to forced vital capacity ratio, but not with age, study location, or analytical method used. Our study suggests that the presence of significant alterations in metabolites within the arginine, transsulfuration, and folic acid pathways can be useful for assessing nitric oxide dysregulation and oxidative stress and identifying novel treatment targets in COPD. (PROSPERO registration number: CRD42023448036.)


Introduction
The global public health and financial burden of chronic obstructive pulmonary disease (COPD) remains unacceptably high despite the availability of different pharmacological and non-pharmacological treatments in this ever-increasing patient group [1][2][3][4][5][6][7].Such challenges have stimulated a significant body of research to better understand the molecular, biochemical, and cellular mechanisms underpinning the pathophysiology of COPD and identify novel druggable targets and therapies [8][9][10][11].Whilst the role of local (airway) and systemic inflammation in COPD is well established, using conventional biomarkers (e.g., C-reactive protein) and specific blood cell types [12][13][14][15], studies have also focused on the dysregulation of the endogenous messenger nitric oxide (NO) and the redox state [16][17][18][19][20][21][22][23][24].The investigation of possible alterations in the NO pathway and redox balance are also important in this context given their involvement in other disease states, some of them, e.g., atherosclerosis and cardiovascular disease, frequently associated with COPD [25][26][27][28][29][30][31].For example, in epidemiological studies, the prevalence of atherosclerotic cardiovascular disease in patients with COPD has been shown to range between 20% and 60% [32][33][34].Furthermore, the coexistence of COPD and cardiovascular disease is associated with poorer quality of life and functional capacity and a higher risk of COPD exacerbations, hospitalizations, and mortality [35][36][37][38].
A significant limitation in the development of analytical platforms for the assessment of NO and biomarkers of oxidative stress in biological samples is represented by the highly reactive nature of these compounds, the relatively short half-life of NO, and the influence of other factors in the assessment of circulating NO metabolites such as nitrite and nitrate [39][40][41][42][43][44][45].Therefore, an alternative approach consists of measuring stable metabolites within metabolic pathways that are closely associated with NO synthesis and oxidative stress.In this context, several metabolites within the arginine, transsulfuration, and folic acid metabolic pathways have been shown to reflect alterations in NO synthesis and/or redox state.Furthermore, these metabolites can be measured in serum or plasma using a wide range of analytical methods for targeted metabolomic analysis, involving the assessment of pre-defined metabolites within specific biochemical pathways (Figure 1) [46][47][48][49][50][51][52][53][54].The arginine pathway includes (a) arginine, a critical amino acid and substrate for several enzymes, e.g., protein arginine methyltransferases (PRMTs), arginase 1 and 2, and NO synthases (NOS) [46,55]; (b) citrulline, the end product of enzymatic reactions catalyzed by NOS and isoform 1 of dimethylarginine dimethylaminohydrolase (DDAH1) [46,56]; (c) the methylated arginine analogues, asymmetric (ADMA) and symmetric (SDMA) methylarginine, which directly (ADMA) or indirectly (SDMA) downregulate NO synthesis [55][56][57][58][59]; and (d) ornithine, the end product of arginase 1 and 2 (Figure 1) [46,60].The transsulfuration pathway regulates sulfur metabolism and redox balance and primarily involves the transfer of sulfur from homocysteine, a highly reactive amino acid derived from the dietary compound, methionine, to cysteine through the intermediate cystathionine, in enzymatic reactions that require vitamin B 6 (Figure 1) [48,61].Finally, the folic acid pathway plays a critical role in regulating several intracellular homeostatic mechanisms that also include the lowering of homocysteine concentrations through the regeneration of methionine in enzymatic reactions that involve vitamin B 12 (Figure 1) [62,63].
Cells 2023, 12, x FOR PEER REVIEW 2 of 27 focused on the dysregulation of the endogenous messenger nitric oxide (NO) and the redox state [16][17][18][19][20][21][22][23][24].The investigation of possible alterations in the NO pathway and redox balance are also important in this context given their involvement in other disease states, some of them, e.g., atherosclerosis and cardiovascular disease, frequently associated with COPD [25][26][27][28][29][30][31].For example, in epidemiological studies, the prevalence of atherosclerotic cardiovascular disease in patients with COPD has been shown to range between 20% and 60% [32][33][34].Furthermore, the coexistence of COPD and cardiovascular disease is associated with poorer quality of life and functional capacity and a higher risk of COPD exacerbations, hospitalizations, and mortality [35][36][37][38].A significant limitation in the development of analytical platforms for the assessment of NO and biomarkers of oxidative stress in biological samples is represented by the highly reactive nature of these compounds, the relatively short half-life of NO, and the influence of other factors in the assessment of circulating NO metabolites such as nitrite and nitrate [39][40][41][42][43][44][45].Therefore, an alternative approach consists of measuring stable metabolites within metabolic pathways that are closely associated with NO synthesis and oxidative stress.In this context, several metabolites within the arginine, transsulfuration, and folic acid metabolic pathways have been shown to reflect alterations in NO synthesis and/or redox state.Furthermore, these metabolites can be measured in serum or plasma using a wide range of analytical methods for targeted metabolomic analysis, involving the assessment of pre-defined metabolites within specific biochemical pathways (Figure 1) [46][47][48][49][50][51][52][53][54].The arginine pathway includes (a) arginine, a critical amino acid and substrate for several enzymes, e.g., protein arginine methyltransferases (PRMTs), arginase 1 and 2, and NO synthases (NOS) [46,55]; (b) citrulline, the end product of enzymatic reactions catalyzed by NOS and isoform 1 of dimethylarginine dimethylaminohydrolase (DDAH1) [46,56]; (c) the methylated arginine analogues, asymmetric (ADMA) and symmetric (SDMA) methylarginine, which directly (ADMA) or indirectly (SDMA) downregulate NO synthesis [55][56][57][58][59]; and (d) ornithine, the end product of arginase 1 and 2 (Figure 1) [46,60].The transsulfuration pathway regulates sulfur metabolism and redox balance and primarily involves the transfer of sulfur from homocysteine, a highly reactive amino acid derived from the dietary compound, methionine, to cysteine through the intermediate cystathionine, in enzymatic reactions that require vitamin B6 (Figure 1) [48,61].Finally, the folic acid pathway plays a critical role in regulating several intracellular homeostatic mechanisms that also include the lowering of homocysteine concentrations through the regeneration of methionine in enzymatic reactions that involve vitamin B12 (Figure 1) [62,63].Importantly, the known associations between the arginine, transsulfuration, and folic acid pathways, vascular homeostasis, and cardiovascular outcomes might also allow investigating the complex interplay between COPD, NO, oxidative stress, and atherosclerotic cardiovascular disease [55,56,58,[64][65][66][67][68][69][70][71][72][73][74].This knowledge would be potentially useful for identifying new therapeutic targets and management approaches in patients with COPD.
We investigated this issue by (a) appraising the available evidence, through a systematic review and meta-analysis, of the association between the circulating concentrations of key metabolites within the arginine, transsulfuration, and folic acid metabolic pathways and COPD, and (b) assessing, where possible, the relationship between the effect size of the observed differences vs. healthy controls and clinical and demographic characteristics.
Two investigators independently screened the abstracts, full-text articles, and relevant references according to the following inclusion criteria: (a) the assessment of homocysteine, cysteine, methionine, vitamin B 6 , vitamin B 12 , folic acid, arginine, ADMA, SDMA, citrulline, or ornithine in plasma or serum, (b) the study of patients with stable COPD and healthy controls using a case-control design, (c) the inclusion of participants ≥18 years, and (d) the availability of full text in English language.The main exclusion criterion was the assessment of patients with acute exacerbations of COPD.The two investigators independently extracted the following variables into an electronic spreadsheet for further analysis: year of publication, first author, study country, participant number, age, male to female ratio, forced expiratory volume in the first second (FEV 1 ), FEV 1 /forced vital capacity (FVC), biological matrix (plasma or serum), and analytical method used.A third investigator was involved in case of disagreement.
The Joanna Briggs Institute Critical Appraisal Checklist was used to assess the risk of bias [75], whereas the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) Working Group system was used to assess the certainty of evidence [76].The Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 statement was followed to present the results [77], and the International Prospective Register of Systematic Reviews was used to register our review (PROSPERO registration number: CRD42023448036).

Statistical Analysis
We created forest plots of standardized mean differences (SMDs) and 95% confidence intervals (CIs) (p-value < 0.05 for statistical significance), and estimated means and standard deviations from medians and interquartile ranges or ranges [78,79], or using the Graph Data Extractor software beta version (San Diego, CA, USA).The heterogeneity of SMD was evaluated using the Q statistic (significance level set at p < 0.10) [80,81].Sensitivity analysis was used to assess the stability of the results [82].The Egger's and Begg's tests and the "trim-and-fill" method were used to assess publication bias [83][84][85].Univariate meta-regression and subgroup analyses investigated associations between the effect size and the following parameters: year of publication, study continent, sample size, age, male to female ratio, FEV 1 , FEV 1 /FVC, biological matrix, and analytical method used.Statistical analyses were performed using Stata 14 (Stata Corp., College Station, TX, USA).

Literature Search
From a total of 1788 articles, we excluded 1759, as they were either duplicates or irrelevant.A full-text revision of the remaining 29 articles led to the exclusion of further eight because they had missing data (n = 2), unsuitable (not case-control) design (n = 4), or included patients with acute exacerbation of COPD (n = 2).The 21 studies included in the final analysis were published between 1998 and 2020 (Figure 2 and Table 1) .There was no disagreement between the two independent investigators; therefore, input from a third investigator was not required.The cross-sectional design of all studies was primarily responsible for the initial low level of certainty given (rating 2, ⊕⊕ ).The risk of bias was low in all studies (Supplementary Table S3) .
Cells 2023, 12, x FOR PEER REVIEW 4 of 27 SMD was evaluated using the Q statistic (significance level set at p < 0.10) [80,81].Sensitivity analysis was used to assess the stability of the results [82].The Egger's and Begg's tests and the "trim-and-fill" method were used to assess publication bias [83][84][85].Univariate meta-regression and subgroup analyses investigated associations between the effect size and the following parameters: year of publication, study continent, sample size, age, male to female ratio, FEV1, FEV1/FVC, biological matrix, and analytical method used.Statistical analyses were performed using Stata 14 (Stata Corp., College Station, TX, USA).

Literature Search
From a total of 1788 articles, we excluded 1759, as they were either duplicates or irrelevant.A full-text revision of the remaining 29 articles led to the exclusion of further eight because they had missing data (n = 2), unsuitable (not case-control) design (n = 4), or included patients with acute exacerbation of COPD (n = 2).The 21 studies included in the final analysis were published between 1998 and 2020 (Figure 2 and Table 1) .There was no disagreement between the two independent investigators; therefore, input from a third investigator was not required.The cross-sectional design of all studies was primarily responsible for the initial low level of certainty given (rating 2, ⊕⊕⊝⊝).The risk of bias was low in all studies (Supplementary Table S3) .

Cysteine
Cysteine was measured plasma in two European studies including a total of 73 COPD patients (mean age: 72 years, 66% males) and 83 healthy controls (mean age: 70 years, 65% males) [87,106].Liquid chromatography with ultraviolet detection was used in one study [87], and capillary electrophoresis with laser-induced fluorescence in the other [106] Cysteine concentrations were significantly higher in COPD patients compared to controls (SMD = 0.34, 95% CI 0.02 to 0.66, p = 0.038; I 2 = 0.0%, p = 0.83; Figure 13).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty remained low (rating 2, ⊕⊕ ) after considering the low risk of bias in all studies, the high but partially explainable heterogeneity, the lack of indirectness, the relatively low imprecision, the moderate effect size, and the lack of publication bias.

Cysteine
Cysteine was measured plasma in two European studies including a total of 73 COPD patients (mean age: 72 years, 66% males) and 83 healthy controls (mean age: 70 years, 65% males) [87,106].Liquid chromatography with ultraviolet detection was used in one study [87], and capillary electrophoresis with laser-induced fluorescence in the other [106].
Cysteine concentrations were significantly higher in COPD patients compared to controls (SMD = 0.34, 95% CI 0.02 to 0.66, p = 0.038; I 2 = 0.0%, p = 0.83; Figure 13).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty remained low (rating 2, ⊕⊕⊝⊝) after considering the low risk of bias in all studies, the high but partially explainable heterogeneity, the lack of indirectness, the relatively low imprecision, the moderate effect size, and the lack of publication bias.

Cysteine
Cysteine was measured plasma in two European studies including a total of 73 COPD patients (mean age: 72 years, 66% males) and 83 healthy controls (mean age: 70 years, 65% males) [87,106].Liquid chromatography with ultraviolet detection was used in one study [87], and capillary electrophoresis with laser-induced fluorescence in the other [106] Cysteine concentrations were significantly higher in COPD patients compared to controls (SMD = 0.34, 95% CI 0.02 to 0.66, p = 0.038; I 2 = 0.0%, p = 0.83; Figure 13).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty was downgraded to very low (rating 1, ⊕ ) after considering the low risk of bias in all studies, the virtually absent heterogeneity, the lack of indirectness, the relatively low imprecision, the relatively small effect size, and the lack of assessment of publication bias (downgrade one level).

Methionine
Two studies investigated plasma methionine in a total of 42 COPD patients and 38 healthy controls [86,88].One study was conducted in Europe [86], and the other in Asia one [88].Liquid chromatography with fluorimetric detection was used in both studies [86,88].
Methionine concentrations were significantly lower in COPD patients compared to controls (SMD = −0.50,95% CI −0.95 to −0.05, p = 0.029; I 2 = 0.0%, p = 0.80; Figure 14).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty was downgraded to very low (rating 1, ⊕⊝⊝⊝) after considering the low risk of bias in all studies, the virtually absent heterogeneity, the lack of indirectness, the relatively low imprecision, the relatively small effect size, and the lack of assessment of publication bias (downgrade one level).

Methionine
Two studies investigated plasma methionine in a total of 42 COPD patients and 38 healthy controls [86,88].One study was conducted in Europe [86], and the other in Asia one [88].Liquid chromatography with fluorimetric detection was used in both studies [86,88].
Methionine concentrations were significantly lower in COPD patients compared to controls (SMD = −0.50,95% CI −0.95 to −0.05, p = 0.029; I 2 = 0.0%, p = 0.80; Figure 14).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty was downgraded to very low (rating 1, ⊕⊝⊝⊝) after considering the low risk of bias in all studies, the virtually absent heterogeneity, the lack of indirectness, the relatively low imprecision, the relatively moderate effect size, and the lack of assessment of publication bias (downgrade one level).
There were non-significant differences in vitamin B12 concentrations between COPD patients and controls (SMD = −0.20,95% CI −0.48 to 0.08, p = 0.16; I 2 = 54.7%,p = 0.11; Figure 15).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty was downgraded to very low (rating 1, ⊕ ) after considering the low risk of bias in all studies, the virtually absent heterogeneity, the lack of indirectness, the relatively low imprecision, the relatively moderate effect size, and the lack of assessment of publication bias (downgrade one level).
There were non-significant differences in vitamin B 12 concentrations between COPD patients and controls (SMD = −0.20,95% CI −0.48 to 0.08, p = 0.16; I 2 = 54.7%,p = 0.11; Figure 15).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.
The level of certainty was downgraded to very low (rating 1, ⊕ ) after considering the low risk of bias in all studies, the moderate heterogeneity, the lack of indirectness, and the lack of assessment of publication bias (downgrade one level).The level of certainty was downgraded to very low (rating 1, ⊕⊝⊝⊝) after considering the low risk of bias in all studies, the moderate heterogeneity, the lack of indirectness, and the lack of assessment of publication bias (downgrade one level).
Folic acid concentrations were significantly lower in COPD patients compared to controls (SMD = −0.37,95% CI −0.65 to −0.09, p = 0.009; I 2 = 0.0%, p = 0.88; Figure 16).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.
Folic acid concentrations were significantly lower in COPD patients compared to controls (SMD = −0.37,95% CI −0.65 to −0.09, p = 0.009; I 2 = 0.0%, p = 0.88; Figure 16).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty was downgraded to very low (rating 1, ⊕⊝⊝⊝) after considering the low risk of bias in all studies, the moderate heterogeneity, the lack of indirectness, and the lack of assessment of publication bias (downgrade one level).
Folic acid concentrations were significantly lower in COPD patients compared to controls (SMD = −0.37,95% CI −0.65 to −0.09, p = 0.009; I 2 = 0.0%, p = 0.88; Figure 16).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty was downgraded to very low (rating 1, ⊕ ) after considering the low risk of bias in all studies, the virtually absent heterogeneity, the lack of indirectness, the relatively low imprecision, the relatively moderate effect size, and the lack of assessment of publication bias (downgrade one level).
There were non-significant between-group differences in arginine concentrations (SMD = 1.53, 95% CI −0.69 to 3.75, p = 0.18; I 2 = 97.7%,p < 0.001; Figure 17).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty was downgraded to very low (rating 1, ⊕⊝⊝⊝) after considering the low risk of bias in all studies, the virtually absent heterogeneity, the lack of indirectness, the relatively low imprecision, the relatively moderate effect size, and the lack of assessment of publication bias (downgrade one level).
There were non-significant between-group differences in arginine concentrations (SMD = 1.53, 95% CI −0.69 to 3.75, p = 0.18; I 2 = 97.7%,p < 0.001; Figure 17).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty was downgraded to extremely low (rating 0, ⊝⊝⊝⊝) after considering the low risk of bias in all studies, the high and unexplained heterogeneity (downgrade one level), the lack of indirectness), and the lack of assessment of publication bias (downgrade one level).
ADMA concentrations were significantly higher in COPD patients compared to controls (SMD = 1.27, 95% CI 0.08 to 2.46, p = 0.037; I 2 = 97.2%,p < 0.001; Figure 18).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty was downgraded to extremely low (rating 0, ) after considering the low risk of bias in all studies, the high and unexplained heterogeneity (downgrade one level), the lack of indirectness), and the lack of assessment of publication bias (downgrade one level).
ADMA concentrations were significantly higher in COPD patients compared to controls (SMD = 1.27, 95% CI 0.08 to 2.46, p = 0.037; I 2 = 97.2%,p < 0.001; Figure 18).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.
The level of certainty was downgraded to very low (rating 1, ⊕ ) after considering the low risk of bias in all studies, the high and unexplained heterogeneity (downgrade one level), the lack of indirectness, the relatively low imprecision, the relatively large effect size (upgrade one level), and the lack of assessment of publication bias (downgrade one level).
SDMA concentrations were significantly higher in COPD patients compared to controls (SMD = 3.94, 95% CI 0.79 to 7.08, p = 0.014; I 2 = 98.1%, p < 0.001; Figure 19).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty was downgraded to very low (rating 1, ⊕ ) after considering the low risk of bias in all studies, the high and unexplained heterogeneity (downgrade one level), the lack of indirectness, the relatively low imprecision, the relatively large effect size (upgrade one level), and the lack of assessment of publication bias (downgrade one level).

Ornithine
Plasma ornithine was measured in three studies including a total of 117 COPD patients (mean age: 69 years) and 82 healthy controls (mean age: 58 years) [86,88,103].Two studies were conducted in Asia [88,103], and one in Europe [86].Liquid chromatography with fluorimetric detection was used in two studies [86,88], and liquid chromatography with ultraviolet detection was used in the remaining one [103].
Ornithine concentrations were significantly higher in COPD patients than controls (SMD = 0.67, 95% CI 0.13 to 1.22, p = 0.015; I 2 = 62.9%, p = 0.067; Figure 20).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.
grade one level), the lack of indirectness, the relatively low imprecision, the relatively large effect size (upgrade one level), and the lack of assessment of publication bias (downgrade one level).

Ornithine
Plasma ornithine was measured in three studies including a total of 117 COPD patients (mean age: 69 years) and 82 healthy controls (mean age: 58 years) [86,88,103].Two studies were conducted in Asia [88,103], and one in Europe [86].Liquid chromatography with fluorimetric detection was used in two studies [86] [88], and liquid chromatography with ultraviolet detection was used in the remaining one [103].
Ornithine concentrations were significantly higher in COPD patients than controls (SMD = 0.67, 95% CI 0.13 to 1.22, p = 0.015; I 2 = 62.9%, p = 0.067; Figure 20).The limited number of studies prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regression and subgroup analyses.The level of certainty was downgraded to very low (rating 1, ⊕⊝⊝⊝) after considering the low risk of bias in all studies, the moderate heterogeneity, the lack of indirectness, the relatively low imprecision, the moderate effect size, and the lack of assessment of publication bias (downgrade one level).
In a study comparing 12 COPD patients (66 ± 2 years) and eight age-matched healthy controls (64 ± 3 years), there were non-significant differences in plasma citrulline concentrations between the two groups (48 ± 6 vs. 54 ± 7 µmol/L) using a liquid chromatography assay with fluorimetric detection [86].The level of certainty was downgraded to very low (rating 1, ⊕ ) after considering the low risk of bias in all studies, the moderate heterogeneity, the lack of indirectness, the relatively low imprecision, the moderate effect size, and the lack of assessment of publication bias (downgrade one level).

Discussion
We observed significant alterations in the circulating concentrations of key metabolites within the arginine, transsulfuration, and folic acid metabolic pathways in COPD using targeted metabolomic analysis.Compared to healthy controls, patients with stable COPD had significantly lower concentrations of methionine and folic acid, and higher concentrations of homocysteine and cysteine.In the context of arginine pathways, COPD was also associated with significant elevations of ADMA, SDMA, and ornithine.Subgroup analysis, which was only possible for studies investigating homocysteine, showed that the SMD of this metabolite was significantly associated with the biological matrix assessed (plasma vs. serum) and the FEV 1 to FVC ratio, but not with age, study location, or analytical method used.
Homocysteine, a highly reactive sulfur-containing amino acid and a metabolite of methionine (Figure 1), has been extensively investigated in view of its capacity to disrupt vascular homeostasis through the inhibition of NO synthesis, endothelial dysfunction, and stimulation of pro-inflammatory and pro-oxidative pathways in the vascular wall and systemically [68,[107][108][109][110][111][112][113][114][115].Not surprisingly, higher circulating homocysteine concentrations have been associated with an increased risk of cardiovascular morbidity and mortality in several observational studies [68,116,117].Notably, homocysteine can also inhibit DDAH1 with a consequent accumulation of ADMA [109], whereas folic acid and vitamin B 12 stimulate the conversion of homocysteine into methionine [68], with consequent homocysteine lowering.These effects further highlight the complex interplay between the arginine, transsulfuration, and folic acid metabolic pathways (Table 1).
The results of our systematic review and meta-analysis, particularly the increased circulating concentrations of homocysteine and ADMA, and the reduced concentrations of folic acid and methionine suggests a significant dysregulation of these pathways in COPD.Such dysregulation would manifest biologically as an impaired synthesis of NO via ADMA accumulation, a pro-oxidative state via homocysteine accumulation, and an overall pro-atherosclerotic state.Furthermore, epidemiological studies have reported that higher ADMA concentrations are independently associated with a significant reduction in FEV 1 and FVC [118].Similar negative associations with FEV 1 and FVC have been reported specifically in healthy smokers [119].In further support of these observations, a study has also reported that patients with COPD have a significantly lower dietary intake of folic acid compared to healthy controls (231 ± 90 vs. 261 ± 110 µg/day, p < 0.001) [120].Notably, in this study, COPD patients in the upper quartile of folic acid intake had significantly lower breathlessness and higher FEV 1 and FVC values compared to patients in the bottom quartile.In a more recent nationwide survey of COPD patients, folic acid concentrations were positively associated with FEV 1 and FVC values, particularly in males and in current smokers [121].Given the well-known homocysteine-lowering effects of folic acid supplementation [68,71,112], and the emerging evidence of additional lowering effects on circulating ADMA [122][123][124][125][126], further studies are warranted to determine whether folic acid supplementation, with or without vitamin B 12 , can improve symptoms, lung function, and clinical outcomes in patients with COPD.
The observed increases in circulating SDMA and cysteine in COPD are intriguing.Like ADMA, SDMA is derived from the methylation of arginine residues in proteins by PRMT 2 [127,128] (Figure 1).However, unlike ADMA, SDMA does not directly inhibit NOS nor is metabolized by DDAH1 and is eliminated in the urine unchanged [56,58].In experimental studies, SDMA has been shown to indirectly reduce NO availability by favoring the uncoupling of NOS and by competing with the transport of the essential NOS substrate arginine [129][130][131].The relatively high prevalence of chronic kidney disease in patients with COPD might potentially account for the reduced renal elimination and consequent accumulation of SDMA in this group [132][133][134].However, recent studies have also reported an association between COPD and PRMTs.For example, an increased expression of PRMT 7, which has also been demonstrated to synthesize SDMA [135][136][137], has been observed in lung tissue macrophages of patients with COPD.Furthermore, a reduced expression of PRMT 7 in mice models of COPD was associated with a reduction in markers of lung injury [137].The increase in cysteine concentrations in COPD is counterintuitive, given that this thiol is essential for protein synthesis, exerts antioxidant effects, and is a precursor of the major antioxidant glutathione and another metabolite with antioxidant effects, taurine [48,[138][139][140].Additional research is required to confirm these findings and elucidate the mechanisms involved in cysteine elevations, including a selective dysregulation of enzymes responsible for its synthesis and degradation [48].
Another interesting observation in our systematic review and meta-analysis was the higher concentration of circulating ornithine in patients with stable COPD compared to healthy controls.As previously described (Figure 1), ornithine is the end product of the arginase 1 and 2 enzymes [60].Therefore, an increase in ornithine concentrations is suggestive of an increased expression and/or activity of arginase which, in turn, reduces the availability of arginine as a NOS substrate for the synthesis of NO.However, this theory has been recently challenged by an elegant in vitro study investigating enzyme kinetics, which reported that the competition between arginase and NOS for the same substrate, arginine, does not occur in the presence of a maintained supply of extracellular arginine, which more adequately reflects the cellular physiological conditions.In the same study, the investigators reported that alterations in arginine transport and/or protein synthesis are more likely to influence NOS activity [141].Pending additional studies investigating the possible involvement of arginase on NO synthesis, arginase upregulation has been reported in experimental models of COPD and clinical studies.For example, mice exposed to cigarette smoking for 13 weeks showed a significant increase in the expression of arginase [142].Similar smoking-mediated increases in arginase expression have been observed in rabbits, with a concomitant reduction in NOS expression and activity [143].Furthermore, treatment with arginase inhibitors significantly suppressed bronchial reactivity in patients with COPD [144].An increased arginase activity has also been reported in platelets and erythrocytes in this group [145].Pending confirmatory studies, this observation suggests that pharmacological strategies downregulating arginase might provide beneficial effects in COPD, independently of NO synthesis [146][147][148][149].
Our study had several strengths, including the comprehensive assessment of arginine, transsulfuration, and folic acid metabolomics in stable COPD and the robust evaluation of the risk of bias and the certainty of evidence for each studied metabolite.Limitations included the small group of selected studies for most metabolites, with the exception of homocysteine, which prevented sensitivity analysis, the assessment of publication bias, and the conduct of meta-regressions and subgroup analyses to investigate associations between the effect size and several clinical and demographic variables, e.g., age, sex, and markers of inflammation, and to identify possible sources of heterogeneity.Further studies are also necessary to investigate the potential pathophysiological role of citrulline and vitamin B 6 , given that our systematic search identified only one relevant study for each metabolite.Another significant limitation was the paucity of data reported in the selected studies regarding specific comorbidities, e.g., neurological and cardiovascular disease states, dietary patterns, and medications, factors which could also affect the concentrations of the studied metabolites [55,56,68,126,[150][151][152][153][154][155][156].At the same time, however, the assessment of the concentrations of folic acid, vitamin B 6 , and B 12 may indirectly reflect dietary behaviours given their associations with specific food sources [157,158].

Conclusions
Our study showed significant alterations in the circulating concentrations of methionine, homocysteine, and cysteine (transsulfuration pathway), folic acid (folic acid pathway), and ADMA, SDMA, and ornithine (arginine pathway) in COPD.These alterations are suggestive of impaired NO synthesis and redox balance and may also explain the frequent occurrence of specific comorbidities, particularly atherosclerotic cardiovascular disease, in this patient group.Further research is warranted to confirm these findings, to investigate further associations between these metabolites and age, sex, markers of inflammation, specific comorbidities, dietary patterns, and medications, and to assess the effects of ADMA/homocysteine-lowering therapies and arginase inhibitors on lung function, symptom burden, disease progression, and mortality in COPD.

Figure 8 .
Figure 8. Forest plot of studies investigating homocysteine concentrations in COPD patients and controls according to analytical method [87,89-93,104,106].

Figure 9 .
Figure 9. Forest plot of studies investigating homocysteine concentrations in COPD patients and controls according to the detection method used with liquid chromatography [87,89,91,92].

Figure 9 . 27 Figure 10 .
Figure 9. Forest plot of studies investigating homocysteine concentrations in COPD patients and controls according to the detection method used with liquid chromatography [87,89,91,92].Cells 2023, 12, x FOR PEER REVIEW 11 of 27

Figure 13 .
Figure 13.Forest plot of studies investigating cysteine concentrations in COPD patients and controls [87,106].

Figure 13 .
Figure 13.Forest plot of studies investigating cysteine concentrations in COPD patients and controls [87,106].

Figure 13 .
Figure 13.Forest plot of studies investigating cysteine concentrations in COPD patients and controls [87,106].
Legend: NR, not reported; M, male; F, female; COPD, chronic obstructive pulmonary disease; ADMA, asymmetric dimethylarginine; SDMA, symmetric dimethylarginine.The concentration of homocysteine, cysteine, methionine, arginine, citrulline, SDMA, and ornithine is expressed in µmol/L.The concentration of ADMA is expressed in µmol/L or ng/mL.The concentration of vitamin B 6 and folic acid is expressed in ng/mL.The concentration of vitamin B 12 is expressed in pg/mL.