Metabolomics of plasma is a promising approach for the investigation of uveitis pathogenesis. In the present study, we found 18 metabolites and 7 metabolites significantly altered in positive and negative modes, respectively. Among them, there have 19 metabolites increased and 6 metabolites decreased in AAU compare with the control group. We further identified two pathways including D-Glutamine and D-glutamate metabolism and Histidine metabolism. ROC results for diagnostic biomarkers showed that 16 altered metabolites have an AUC > 0.7, which including Glycyltyrosine showed AUC = 0.991. Previous metabolomics of plasma of AAU patients showed involvement of amino acids, carbohydrate, and lipid metabolism based on LC-MS [20], and partially consistent our observations in plasma from AAU.
25 metabolites, which were selected from 483 identified metabolites from AAU and control plasma, were considered as potential biomarkers. Among these different metabolites, several amino acids and lipids were interesting to investigate the exact function of each of them in AAU pathogenesis. PCA model was not showed a good distinction, while the OPLS-DA model showed a clear difference between groups AAU and control. The OPLS-DA model in our study is capable of explaining 96.8% and predicting 95.5% in positive and explaining 97.8% and predicting 95.4% in the negative mode of all metabolomics data and permutation test have verified the OPLS-DA models were properly. Our OPLS-DA model indicated that the strength of modeling and prediction by their good parameters.
There have been some other metabolomic studies of uveitis based on a different platform and different sample types, like aqueous humor and vitreous humor. Several metabolites related to the arginase pathway were significantly increased in lens-induced uveitis (LIU) and chronic uveitis (CU), which were detected through NMR from 42 vitreous samples [28]. Metabolomics of synovial fluid from 6 patients with Behcet’s disease (BD) and 18 patients with seronegative arthritis (SNA) by using GC-MS found significantly metabolic differences and potential metabolic biomarkers for discriminating BD from SNA [29], and they first investigate the characteristic of serum metabolic profiles and potential metabolic biomarkers for diagnosis of BD [30]. Twenty-four metabolites differentially expressed between Vogt-Koyanagi-Harada (VKH) and healthy controls, and D-mannose, stearic acid, sarcosine, and L-lysine were recognized as potential biomarkers based on LC-MS [31] and 24 metabolites showed significantly difference between healthy control and uveitis, which include BD, VKH and sarcoidosis [32]. Fatty acids and amino acids were the two most increased categories of altered metabolites when comparing VKH, BD, and control groups [33]. In our study, we found 11 amino acids and 7 lipids altered metabolites, which is partly consistent with the previous study. Sample of plasma including lots of physiological information of our body changed, which is a good source for metabolomics analysis. We believe the large sample size and high-resolution mass spectrometry used will change the attitude in uveitis studies.
Lipids, including phosphatidylcholines (PCs), have been investigated as potential biomarkers for diagnosis of some diseases, such as major depressive disorder [34], ovarian cancer [35], multiple sclerosis [36], and also involved in fatty liver fibrosis [37], colorectal liver metastasis [38]. Furthermore, PCs have been suggested as a predictive metabolite corresponding with progressive nephropathy [39] and early kidney function declines rapidly in patients with type 1 diabetes [40]. Increased PCs were found in Glucokinase-maturity onset diabetes of the young, which represents a rare genetic disorder due to mutation in the glucokinase (GCK) gene [41]. Recently, PCs showed the strongest correlation with the eye score of NIH, which indicates metabolic dysregulation of tears in ocular chronic graft-versus-host disease [42]. Consistent with the previous study, there have been several PCs changed in AAU compared with the control group, which lists in Table 2 in our study. Supplementary Fig. 3 showed that the level of PC (20:2(11Z,14Z)/15:0), PC (22:2(13Z,16Z)/16:1(9Z)) and PC (16:1(9Z)/20:0) in AAU were decreased, which suggest this may be related with the hyperinflammatory status of AAU and these PCs’ function in AAU need to be further studied.
Amino acids in our body have a vital role in metabolic and catabolic, which can synthesize protein and affect the immune system. A previous study showed several amino acids were discriminate Posner-Schlossman syndrome (PSS) with control [43]. Recently, a multi-omics study, which combine proteomics and metabolomics, showed amino acids metabolic pathways have an vital role in the pathogenesis of VKH using sweat samples [44]. In our study, several amino acids were altered in AAU compared with control, which lists in Table 2. Supplementary Fig. 3 showed that Glycyltyrosine, Pyroglutamic acid, L-Aspartyl-4-phosphate, Brassica oleracea Alkaloid, L-Glutamic acid, gamma-Glutamyl-S-methylcysteine sulfoxide, Diaminopimelic acid, 3-Aminoisobutanoic acid, and Glycylproline increased in AAU and L-Histidine, Tryptophyl-Glutamine decreased in AAU than in the control group. These inconsistent results may be due to heterogeneity of patients and different platform of metabolites detection.
Our study has uncovered several interesting findings, but still has some limitations that might cause bias, especially no health control and retrospective design. Heterogeneity of the participants, which including environmental factors, dietary habits, genetic factors, body mass index (BMI), daily activity, disease status, and the plasma across individuals also confused results. These different kinds of difficulties can be solved by investigating larger sample populations or different subtypes of AAU (active and inactive) or by controlling for important confounders of the study population (such as daily activity, BMI, diet information, medication). Results from plasma metabolomics data have some limitations because it circulating in our body that can be influenced by our immune status. Therefore, combine the samples from the eye (tears, aqueous humor, vitreous humor) of patients will be better for biomarkers discovery. Multicenter cohort and targeted metabolomics validation will be better for these results.
In conclusion, we found specific plasma metabolic profiles for AAU. Altered amino acids and lipids expression in plasma could play a vital role in the pathogenesis of the intraocular inflammation in this disease. Several changed metabolites, especially Glycyltyrosine, have a potential biomarker for AAU diagnosis. The changed metabolic profile can help us to further understand the AAU pathophysiology. Further studies are still needed to explore the function of these metabolites and pathways in animal model of AAU.
Summary Box
What was known before
What this study adds
-
Plasma metabolic signature from larger sample size distinguish the AAU from control.
-
Several changed metabolites, especially Glycyltyrosine, show great diagnostic power for acute anterior uveitis.