Metabolomic Profiling of Ocular Fluid Identifies Host Response Factors for Risk Stratification in Bacterial Endophthalmitis

Naren Shetty; Padmamalini Mahendradas; Ramaraj Kannan; Samayitree Das; Priyanka Sathe; Rohit Shetty; Arkasubhra Ghosh; Vrushali Deshpande

doi:10.1167/iovs.66.5.27

. 2025 May 22;66(5):27. doi: 10.1167/iovs.66.5.27

Metabolomic Profiling of Ocular Fluid Identifies Host Response Factors for Risk Stratification in Bacterial Endophthalmitis

Naren Shetty ¹, Padmamalini Mahendradas ², Ramaraj Kannan ³, Samayitree Das ³, Priyanka Sathe ¹, Rohit Shetty ⁴, Arkasubhra Ghosh ^3,^✉, Vrushali Deshpande ^3,^✉

PMCID: PMC12110596 PMID: 40402515

Abstract

Purpose

This study aimed to identify metabolite biomarkers in ocular fluid that can serve as indicators of infection severity, guide risk stratification, and monitor the host response to bacterial endophthalmitis.

Methods

We performed untargeted metabolomics to profile ocular fluids from 26 patients diagnosed with bacterial endophthalmitis and 17 control samples. Multivariate data analysis and pathway enrichment of metabolites were performed using the MetaboAnalyst R package.

Results

The volcano plot analysis revealed significantly altered m/z values (log fold change > 2; P < 0.05), with 1824 m/z values upregulated and 3854 m/z values downregulated species. Notably, the levels of methyl linolenate significantly decreased from grade 2 to grade 5, whereas no significant alterations were observed in grade 1 compared with controls. The peak intensities of palmitoleamide also showed a decrease from grade 2 to grade 5 compared with controls. Conversely, myristoleoyl carnitine exhibited significantly increased levels in both grade 2 and grade 5. Methyl linolenate (AUC, 0.885), palmitoleamide (AUC, 0.885), and myristoleoyl carnitine (AUC, 0.869) were identified as the three metabolites with the highest classification capacity for bacterial endophthalmitis in tear samples.

Conclusions

This report is the first of a set of metabolite biomarkers in ocular fluid that correlate with the severity of bacterial endophthalmitis. These biomarkers have the potential to aid in risk stratification and personalized treatment approaches. The metabolomic profiles generated in this study provide insight into the host response mechanisms and the pathophysiology of the infection in human eyes.

Keywords: bacterial endophthalmitis, metabolomics, tear biomarker, infection, inflammation

Endophthalmitis is a vision-threatening condition, which is characterized by inflammation within the intraocular chambers and affecting the composition of the vitreous and aqueous fluids.¹ Based on the route of transmission of the infection, endophthalmitis can be classified as exogenous or endogenous. In endogenous endophthalmitis, infection originates from within the body, can be caused by bacteria or fungi, and spreads hematogenously from a different point of infection and accounts for approximately 2% to 8% of all cases of endophthalmitis.² Exogenous endophthalmitis occurs when infectious agents (exogenous microorganisms) are introduced into the eye mostly during postoperative complications or after ocular trauma. The common symptoms of endophthalmitis include redness, swelling, pain, and reduced vision, irrespective of the route of transmission.³ Most severe infections are characterized clinically with alterations in red fundus reflex and significant loss of vision at the time of presentation.⁴ Hypopyon is the classic sign observed in almost three-fourths of patients at the time of presentation. Progressive vitritis may lead to perforation, corneal infiltration, and deformed orbital structures, resulting in panophthalmitis.⁵ In severe cases, vitrectomy (24.1%) and evisceration/exenteration (11.1%) is performed.⁶

The management of bacterial endophthalmitis is done with repeated intravitreal injections of antibiotics like vancomycin (1.0 mg/0.1 mL), ceftazidime (2.25 mg/0.1 mL), or amikacin (0.4 mg/0.1 mL), as well as systemic antibiotics along with topical antibiotics. Intravitreal amphotericin-B (0.005 mg/0.1 mL) or voriconazole (0.1 mg/0.2 mL), along with systemic antifungal agents, are used in cases of fungal endophthalmitis.⁷^,⁸ The anterior and vitreous chamber is separated from the systemic circulation by the blood–ocular barrier, rendering a challenge of penetration for systemic antimicrobial and anti-inflammatory molecules.⁹ The intraocular inflammation may be exacerbated owing to the ongoing tissue damage, as well as the use of antibiotics. The excessive inflammation may lead to photoreceptor or retinal damage, which may cause permanent vision loss. Vitrectomy, a debulking of the vitreous along with the infected organisms, is an alternate management therapy that helps to improve antibiotic diffusion.⁸^,¹⁰^,¹¹ However, rapid progression, subretinal abscess, and retinal detachment are further complications that may worsen the visual outcome.¹²

Further, even when cluster cases of endophthalmitis occur owing to an infected agent in the surgical process, not every patient contracts the infection. This finding suggests that there may be host susceptibility or resistance factors within the eye that can dictate the occurrence of infection. Thus, a detailed understanding of the intraocular host response and its early recognition is necessary for effective treatment and improving visual outcome. Most patients present in the advanced disease state with symptoms of pain and blurry vision.¹⁰ Early intervention in terms of immediate onset of the appropriate treatment is crucial to prevent further damage. Although culture and antibiotic sensitivity testing from the ocular fluids and PCR testing of bacterial 16S rDNA is the gold standard to diagnose bacterial endophthalmitis, it has its drawbacks, including long processing times (≤48 hours) and the need for specialized equipment and species-specific protocols.

The diagnosis of endophthalmitis is based on clinical presentation; however, in the immediate postoperative stage it is challenging to differentiate between toxic anterior segment and endophthalmitis.¹³^,¹⁴ In a single visit, it will be difficult to confirm the diagnosis and multiple clinic visits may be required to confirm the diagnosis. Diagnostic vitrectomy followed by culture and histological examination are important to confirm the presence of a specific etiology.¹⁵^,¹⁶ To address this issue, we investigated the metabolite factors involved in the bacterial endophthalmitis cases. We hypothesized that a tear-based biomarker identified from bacterial endophthalmitis cases would prove useful in risk stratification and in understanding the host response mechanisms and pathophysiology of the infection. This study identified a set of metabolic biomarkers in tears that reflect the severity of bacterial endophthalmitis.

Materials and Methods

Patient Selection

All patients suffering from endophthalmitis of any origin, that is, postoperative, traumatic, endogenous, and post-COVID endophthalmitis were included in the study. Patients with any corneal/scleral infections were excluded from the study to ensure there was no secondary microbial infection interference in the study. All cases with probable endophthalmitis (n = 26) were enrolled for the study after receiving written, informed consent. This study was approved by the Institutional Ethics Committee of Narayana Nethralaya, Bangalore, India (ECR/187/Inst/Kar/2013/RR-19), for the entire cohort study. Patients with no report of ocular infection were included as controls. Individuals who underwent vitrectomy for noninfectious retinal disorders but had noninflamed eyes were considered as the control group (n = 17). Patients with Noninfectious uveitis were also enrolled in the study (n = 5). Only tears were collected from the noninfectious uveitis patient group. All enrolled patients underwent complete ophthalmic and systemic evaluations. Patients with hazy fundus views were taken for B scan ultrasound examination. All samples were collected before treatment. Treatment of the patients began with topical (moxifloxacin 0.5% eye drop) and systemic antibiotics (ciprofloxacin, doxycycline, nitroglycerin tablet); topical prednisolone acetate 1%; topical homatropine 2%; intravitreal vancomycin, and ceftazidime/dexamethasone (half dose). Pars plana vitrectomy was only performed in selected cases where significant media haze was present with vitreous exudates to debulk the infection load. Based on fundus imaging of the retinal vessels and visibility status, patients were stratified into different severity grades (1–5). The patient demographic and grade-wise clinical details are provided in Figure 1 and Table.

Figure 1. — (A) Fundus image of a normal eye. (B) Fundus photograph of a 19-year-old female right eye showing grade 2 media haze with vitreous exudates in bacterial endophthalmitis. (C) Fundus photograph shows grade 3 media haze in a case of bacterial endophthalmitis. (D) Fundus image of grade 4 media haze with snowball opacities in a case of bacterial endophthalmitis.

Table.

Demographic Details of the Patient Cohort

Variables	Control (n)	Sample (n)
Gender
Male	10	13
Female	7	13
Age, years (range)	26–84	13–81
Etiology
Bacterial endophthalmitis	0	26
Noninfectious uveitis	5	0
Sample
Tear	17	26
Aqueous humor	6	12
Vitreous humor	2	2
Anterior chamber
No hypopyon, fibrin, flare and inflammatory cells		11
Hypopyon, fibrin, flare (2+), inflammatory cells (2+)		6
Hypopyon, fibrin, flare (3+), inflammatory cells (3+)		7
Hypopyon, fibrin, flare (4+), inflammatory cells (4+)		4
Vitreous chamber
No inflammation		11
Vitreous haze		6
Inflammatory cells (3+)		7
No view		4
Cornea
No infiltration of inflammatory cells		11
Presence of Inflammatory cells, DM folds, keratitis precipitate		17
Retina
Normal view of the retina		11
Retinal vessels		17

Open in a new tab

Sample Collection

All samples were collected during the active phase of infection, either at the time of clinical presentation or within a few days after diagnosis. Sample collection was performed before the initiation of any treatment to ensure an unaltered biological profile. Tears were collected in a Schirmer's strip with strict sterile precautions using a no-touch technique and stored at −80°C in a sterile Eppendorf tube. Anterior chamber paracentesis with strict sterile precautions was carried out to collect 0.2 mL of aqueous humor using no-touch techniques. We transferred 50 µL of aqueous humor to a new tube, sealed and labelled it, and stored at −80°C. Vitreous humor was collected under aseptic precautions in the OT with a vitreous cutter during diagnostic or therapeutic vitrectomy. We transferred 50 µL of vitreous humor to a new tube for metabolic profiling and stored at −80°C.

Metabolomic Profiling

Sample Processing for Liquid Chromatography Coupled With a Tandem Mass Spectrometer

Schirmer's strips were cut into small pieces and 200 µL of 9:1 methanol: water was added. Aqueous humor and vitreous humor samples were mixed with 1:6 volume of 1:1 methanol: ethanol (100%). All three sample types with the respective solvents were then vortexed vigorously for 2 minutes and kept in a thermomixer at 4°C for 30 minutes at 1250 rpm. The suspension was centrifuged at 14,000 rpm for 10 minutes at 4°C. Supernatant was collected in a fresh tube and dried in a vacuum centrifuge. The dried pellet was resuspended with 80 µL of 50% ACN. Global metabolomic profiling was done by Exion LC UHPLC coupled to a Triple TOF SCIEX 5600+ mass spectrometer.

The chromatographic separation was achieved using Kinetex 2.6µm C18 column (100Å, 100 × 4.6 mm; Product No: 00D-4462-E0). The samples were analyzed in both positive and negative modes of ionization. For positive mode, water (Product No: LC365) with 0.1% formic acid and acetonitrile (product no: 34967) with 0.1% formic acid (product no: AAB-A117-50) were used as mobile phase. For negative mode water with 0.5mM ammonium formate (product no: 78314) and acetonitrile were used as the mobile phase. A 35-minute binary gradient mode with the flow rate of 0.4 mL/min was set for the sample elution as follows: 0 minute, 5% B; 4 minutes 50% B; 20 minutes 90% B; 23 minutes 90% B; 26 minutes 100% B; 27 minutes 100% B; 33 minutes 5% B; 35 minutes 5% B. The column temperature and the autosampler temperature were set as 30°C and 4°C, respectively. The ESI capillary temperature was maintained at 550°C. The DDA acquisition with the cycle time of 1 second and accumulation time of 0.25 seconds were set.

Preprocessing, annotation, and statistical analysis of metabolomic data. The metabolomics raw data were acquired from liquid chromatography triple time of flight mass spectrometry in a .wiff format. The datasets were fed into the Peak View software to acquire the highest area under the peak time duration. Marker view v1.3.1 was used to obtain the mass-to-charge ratio (m/z), retention time and peak intensity. The peaks obtained between 1 minute and 35 minutes were scanned using Marker view v1.3.1 with the minimum intensity count of 5. Approximate peak width was set as 0.4 seconds with the mass tolerance of 10 ppm. Retention time tolerance was set to 2 minutes. The Omics craft tools (Metaboquest) were used to annotate isotope adducts, peaks, and fragments (http://tools.omicscraft.com/MetaboQuest/). MetaboAnalyst R package was used to perform multivariate data analysis of metabolites.¹⁷ The samples were normalized by median and the pareto data scaling was used. Supervised partial least-squares discriminant analysis (PLS-DA) method was performed to identify clusters to locate metabolites across various grades of different sample types. The variable importance of projection (VIP) score score was measured as a coefficient to select each variable. To understand the importance of each metabolite the nonparametric Kruskal–Wallis statistical test was performed. To classify the key altered metabolites random forest and heat map analysis was performed using the R package “caret” of MetaboAnalyst R.¹⁷ Pathway enrichment analysis of the important metabolites was performed using Functional analysis (one factor) from MetaboAnalyst R.

Results

To identify the alteration of the key metabolites and pathways in bacterial endophthalmitis, we performed metabolomic profiling of tears and aqueous and vitreous endophthalmitis samples. The schematic representation of the workflow is shown in Figure 2A. The Venn diagram showing the number of common and unique m/z values identified in tears, aqueous humor and vitreous samples is shown in Figure 2B. Bacterial endophthalmitis samples were divided into different grades (grades 1–5) concerning the severity of the infection (Fig. 1). Our data indicate a correlation between different severity grades of bacterial endophthalmitis and clinical parameters such as visual acuity, retinal vessel visibility in the fundus, and the number of inflammatory cells. PLS-DA was performed to identify differentially produced metabolites in tear, aqueous humor and vitreous humor samples in bacterial endophthalmitis patients with respect to controls.

Figure 2. — (A) Schematic representation of the experimental workflow. (B) Venn diagram showing unique and common *m/z* values in tears, vitreous humor, and aqueous humor of bacterial endophthalmitis patients.

Altered Levels of Metabolites in Tears of Bacterial Endophthalmitis Patients

The volcano plot in Figure 3A shows the significantly altered m/z values (log fold change >2; P < 0.05), with 1824 m/z to be upregulated and 3854 m/z to be downregulated. These m/z values were annotated against the Human Metabolome database and yielded 122 upregulated and 1014 downregulated metabolites. We performed PLS-DA to identify altered levels of metabolites across different grades with respect to controls with variance of (Fig. 3B). The variable importance in projection (VIP) scores of top 25 m/z is listed, with 2-chloro-1-[(2S,3S,5S,10S,13S)-3-hydroxy-10,13-dimethyl-2-morpholino-2,3,4,5,6,7,8,9,11,12,14,15,16,17-tetradecahydro-1H-cyclopentaa]phenanthrene-17-yl], 4-ethyl-2-octylthiazole, estreptoquinasa, oxendolone, and 9-octadecen-1-amine, being the top five metabolites, respectively (Fig. 3C). To address the variability among different experimental groups, the classification of these top altered metabolites was further validated using random forest classification model, which showed a prediction error rate to be less than 6% (Fig. 3D). The heat map representation showed the altered levels of metabolites across different grades as compared with the control and uveitis groups (Fig. 3E).

Pathway enrichment analysis showed pathways for the top altered metabolites (Supplementary Table S1), with the most enriched ones being glycerophospholipid metabolism, pentose phosphate pathway, glycine, threonine, and serine metabolism (Fig. 3F).

We analyzed differences in the peak intensities of individual values of six metabolites across different groups involved in lipid–fatty acid metabolism (Fig. 4). Myristoleoyl carnitine is a human C14 acylcarnitine involved in beta-oxidation of long chain fatty acids. L-acylcarnitines activate classical proinflammatory signaling pathways.¹⁸ In tears, myristoleoyl carnitine showed a significant increase in the levels in grade 4 and grade 5 disease. In grades 1 and 2, the levels were not altered significantly as compared with controls (Fig. 4A). The metabolite 2,4,6 tri-tert-butylphenol, lipid class molecule level significantly decreased in grade 3, grade 4, and noninfectious uveitis groups (Fig. 4B). The metabolite 9-octadecenamide peak intensities showed decrease from grade 2 to grade 5 as compared with controls and grade 1. This metabolite was not detected in the uveitis noninfectious group. There was no significant difference observed within grade 2 to grade 5 (Fig. 4C). The methyl linolenate an essential lipid, whose levels decreased from grade 2 to grade 5 as compared with controls and showed highest levels in grade 1 (Fig. 4D). This metabolite was not detected in uveitis noninfectious group. Higher levels of 5-alpha pregnane 3,20-diol, gluco/mineralocorticoids were observed in grade 1 as compared with control. Their levels further decreased in severe infection (grade 2 to grade 5) (Fig. 4E). Palmitoleamide levels significantly decreased in grade 2, grade 4, and grade 5 (Fig. 4F). These results confirmed the altered levels of metabolites in tears during active bacterial infection.

Figure 4. — Altered metabolite levels in tears across different severity grades of bacterial endophthalmitis. Tear (A–B) Bar graphs show peak intensity on the y axis and grades of bacterial endophthalmitis on the x axis for myristoleolyl carnitine, 2,4,6-tri-tert-butyl phenol, 9-octadecenamide, methyl linolenate, 5-alpha-pregnane-3apha 20alpha diol, and palmitoleamide metabolites, respectively. Metabolite levels with significant differences as determined using the nonparametric Kruskal–Wallis test between the different grades are marked with asterisks (**P < 0.05, ***P < 0.001, ****P < 0.0001). Error bars indicate SD obtained from four technical replicates of each biological replicates (n = 48/control = 17, bacterial endophthalmitis = 26, grade 1 = 1, grade 2 = 8, grade 3 = 7, grade 4 = 7, grade 5 = 3, noninfectious uveitis = 5), respectively.

Metabolites Altered in the Aqueous Humor of Bacterial Endophthalmitis Patients

The results of the metabolomic analysis represented in volcano plot shows the significantly altered m/z values (log fold change >2; P < 0.05), with 1733 m/z to be upregulated and 1131 m/z to be downregulated (Fig. 5A). We performed PLS-DA to identify altered levels of metabolites across different grades with respect to controls with a variance of 54% (Fig. 5B). The VIP scores of top 20 metabolites is listed, with 13-docosenamide, L-glutathione, eicosanoic acid, -11-oxo-, methyl ester, and goyaglycoside C being the top metabolites, respectively (Fig. 5C). Owing to the high variability among the different experimental groups, the classification of these top altered metabolites was further validated using random forest classification model, which showed a prediction error rate to be less than 8% (Fig. 5D). The heat map representation showed the altered levels of metabolites across different grades as compared with the control (Fig. 5E). Pathway enrichment analysis showed 42 pathways (Supplementary Table S2), the most enriched ones being phytanic acid peroxisomal oxidation, fatty acid activation, glycerophospholipid metabolism, de novo fatty acid biosynthesis, fructose, and mannose metabolism (Fig. 5F).

Figure 5. — Metabolomic profiling reveals altered metabolite levels in aqueous humor during bacterial endophthalmitis. (A) Volcano plot of aqueous metabolome comparing bacterial endophthalmitis vs. controls. Cutoff for P value is <0.05; log fold change (infected/noninfected) cutoff is >1 or <−1. (B) PLS-DA of untargeted metabolomics data divided based on the severity of the infection indicated the distribution pattern. (C) Top 20 altered metabolites based on VIP scores. (D) Random forest classification model and error identification model to validate the top altered metabolites. Top 15 metabolites that were differentially altered across the various time points, as identified by random forest classification. (E) Heat map of 20 metabolites selected by the PLS-DA VIP score of >1.5 across different grades of endophthalmitis infection. Each row represents a tear sample and each column represents the expression profile of a metabolite across different grades. The changes of fold SD from the overall mean concentration for different years of age are shown in a color-coded way. *Blue color* represents a decrease, and *red color* an increase. (F) Metabolic pathway analysis plot created using MetaboAnalyst 3.0. Plots depict several metabolic pathway alterations induced in case of bacterial infection. The x axis represents the pathway impact value computed from pathway topological analysis, and the y axis is the log of the P value obtained from pathway enrichment analysis. The pathways that were most significantly changed are characterized by both a high log (P) value and high impact value.

Differences in the peak intensity individual values of five metabolites across different groups involved in lipid metabolism were analyzed (Fig. 6). 13-Docosenamide is a chemical compound, which showed a decrease in the levels from grade 1, grade 3, and grade 4, respectively (Fig. 6A). The levels of phosphatidyl glycerol a phospholipid were altered and found to decrease with increasing severity as compared with the controls (Fig. 6B). Goyaglycoside, an exogenous metabolite, is a glycosylated derivative of triterpene sapogenins, which also showed decreased levels in all the grades significantly as compared with the control (Fig. 6C). Eicosanoic acid, -11-oxo-, and methyl ester, also known as arachidonic acid, is a 20-carbon chain saturated fatty acid. Arachidonic acid plays an important role in synthesizing biologically active compounds, such as eicosanoids, thromboxanes, leukotrienes, and prostaglandins, which are among the substrates of cell membrane phospholipids.¹⁹ The levels decreased in all grades with respect to infection severity as compared with the controls (Fig. 6D). 17-Phenyl trinor-prostaglandin E2, classified as a member of prostaglandin levels decreased drastically in grade 4 as compared with control and other grades (Fig. 6E). However, 7-hydroxy-6-methylcoumarin levels increased significantly in grade 3 and grade 4 as compared with controls and grade 1 (Fig. 6F). L-Glutathione levels showed a sequential decrease from grade 1 to grade 4 (Fig. 6G).

Figure 6. — Altered metabolite levels in aqueous humor across different severity grades of bacterial endophthalmitis AH. (A–G) Bar graphs show peak intensity on the y axis and grades of bacterial endophthalmitis on the x axis for 13-docosenamide, phosphatidyl glycerol, goyaglycoside C, ricosanoic acid, -11-oxo-, methyl ester, 17-phenyltrinor-prostaglandin E2, 7-hydroxy 6-methylcoumarin, and L-glutathione metabolites, respectively. Metabolite levels with significant differences as determined using nonparametric Kruskal–Wallis test between the different grades are marked with asterisks (**P < 0.05, ***P < 0.001, ****P < 0.0001). Error bars indicate SD obtained from four technical replicates of each biological replicates (Control = 6; Grade 1 = 2, Grade 2 = 0, Grade 3 = 3, Grade 4 = 3, Grade 5 = 0) respectively.

Metabolites Altered in Vitreous Humor of Bacterial Endophthalmitis Patients

Vitreous humor samples obtained from vitrectomy cases with observable vitreous exudates requiring debulking to reduce infection load. The samples collected were grouped into grades 2 and 4 according to the severity of infection. The volcano plot showed the significantly altered m/z values (log fold change >2; P < 0.05), with 687 m/z to be upregulated and 667 m/z to be downregulated (Fig. 7A). We performed PLS-DA to identify altered levels of metabolites across different grades with respect to controls with variance of 47.5% (Fig. 7B). The VIP scores of top 20 metabolites is listed, with 9-bromo-9, 10, dihydro-N-(2,4-xylyl)-9,10-ethanoanthracene-11, 12, -dicarboximide, myristoleoyl carnitine, and propionamide being the top metabolites, respectively (Fig. 7C). Owing to the high variability among the different experimental groups, the classification of these top altered metabolites was further validated using random forest classification model, which showed a prediction error rate to be less than 4% (Fig. 7D). The heat map representation showed the altered levels of metabolites across different grades as compared with the controls (Fig. 7E).

Figure 7. — Metabolomic profiling reveals altered metabolite levels in vitreous humor during bacterial endophthalmitis. (A) Volcano plot of aqueous metabolome comparing bacterial endophthalmitis vs. controls. Cutoff for the P value is <0.05; log fold change (infected/noninfected) cutoff is >1 or <−1. (B) Partial least squares discriminant analysis (PLS-DA) of untargeted metabolomics data divided based on the severity of the infection indicated the distribution pattern. (C) Top 20 altered metabolites based on VIP scores. (D) Random forest classification model and error identification model to validate the top altered metabolites. Top 15 metabolites that were differentially altered across the various time points, as identified by random forest classification. (E) Heat map of 20 metabolites selected by the PLS-DA VIP score of >1.5 across different grades of endophthalmitis infection. Each row represents a tear sample and each column represents the expression profile of a metabolite across different grades. The changes of fold SD from the overall mean concentration for different years of age are color coded. *Blue color* represents a decrease and *red color* an increase. (F) Metabolic pathway analysis plot created using MetaboAnalyst 3.0. Plots depict several metabolic pathway alterations induced in case of bacterial infection. The x axis represents the pathway impact value computed from pathway topological analysis, and the y axis is the log of the P value obtained from pathway enrichment analysis. The pathways that were most significantly changed are characterized by both a high log (P) value and high impact value.

Pathway enrichment analysis showed 53 pathways (Supplementary Table S3), the most enriched ones being biopterin metabolism, drug metabolism-cytochrome P450, lysine metabolism, sialic acid metabolism, purine metabolism, and glycerophospholipid metabolism (Fig. 7F).

Interestingly, myristoleoyl carnitine was detected in both vitreous humor and in tears (Fig. 8A). The levels increased from grade 2 and grade 4 cases compared with controls. 4-(Tris(trimethylsilyl)) bromobenzene and 3-aminomethyl-pyridinum adenine-dinucleotide levels were increased in only in grade 2 (Figs. 8B and 8D). The metabolite arachidonoyl-2-fluoroethylamide showed significant decrease in the levels in grade 2, but there was no significant change in grade 4 as compared with controls (Fig. 8C). Geranyl diphosphate also showed similar levels (Fig. 8D).

Figure 8. — Altered metabolite levels in vitreous humor across different severity grades of bacterial endophthalmitis. (A–E) Bar graphs show peak intensity on the y axis and grades of bacterial endophthalmitis on the x axis for myristoleolyl carnitine, 4-(tris-(trimethylsilyl)) bromobenzene, arachidonoyl-2-floroethylamide, 3-aminomethyl-pyridinium-adenine-dinucleotide, and geranyl diphosphate metabolites, respectively. Metabolite levels with significant differences as determined using nonparametric Kruskal–Wallis test between the different grades are marked with asterisks (**P < 0.05, ***P < 0.001, ****P < 0.0001). Error bars indicate SD obtained from four technical replicates of each biological replicates (control = 2; grade 1= 0, grade 2 = 1, grade 3 = 0, grade 4 = 1, and grade 5 = 0), respectively.

Validation of the Classification Model in Bacterial Endophthalmitis Diagnosis

To evaluate the potential of these metabolites as biomarkers for bacterial endophthalmitis, receiver operating characteristic curves were plotted for each high score metabolite and the resulting area under the curve (AUC) values were calculated. The three metabolites in tear with the highest AUC values were eventually inculcated into a multivariate signature and the effectiveness of their classification was tested by a linear support machine learning algorithm named as vector machines. Palmitoleamide (AUC, 0.88), myristoleoyl carnitine (AUC, 0.86) and methyl linolenate (AUC, 0.88) were the three metabolites with best classification capacity in tears for bacterial endophthalmitis as compared with normal controls, which was further validated by a multivariate signature (Fig. 9A). The effectiveness of these three discriminating metabolites in tears was explored against noninfectious uveitis yielding AUC values of 0.62 for palmitoleamide, 0.48 for myristoleoyl carnitine, and 0.71 for methyl linolenate (Fig. 9B). The best classification capacity was shown for methyl linolenate metabolite against noninfectious uveitis in their classification. These results suggest that tear fluid may be used to diagnose bacterial endophthalmitis. To determine the effectiveness the three discriminating metabolites tears for risk stratification, different grades of bacterial endophthalmitis against noninfectious uveitis (Supplementary Fig. S2) and control samples (Supplementary Fig. S1) were explored. The effectiveness of these metabolites in tears was also shown between different grades of severity (Supplementary Fig. S3). The best classification capacity was shown by myristoleoyl carnitine AUC values (0.92 grade 2, 0.88 grade 3, 1.00 grade 4, and 0.93 grade 5) across severity grades of bacterial endophthalmitis when compared with control samples. Moreover, myristoleoyl carnitine showed the best classification capacity AUC values (0.73 grade 2, 0.80 grade 3, 0.63 grade 4, and 0.92 grade 5) and methyl linolenate showed AUC values (0.78 grade 3 and 0.94 grade 5) across severity grades of bacterial endophthalmitis when compared with uveitis samples. The best classification capacity was shown by myristoleoyl carnitine AUC values (0.6 grade 2 vs. grade 3, 0.73 grade 3 vs. grade 4, and 0.92 grade 4 vs. grade 5), methyl linolenate AUC values (0.68 grade 2 vs. grade 3, and 0.83 grade 4 vs. grade 5) and palmitoleamide AUC values (0.61 grade 2 vs. grade 3, and 0.62 grade 4 vs. grade 5).

Figure 9. — Biomarker analysis. Box plots and receiver operating characteristic curve (ROC) AUCs curves were determined for selected metabolite in tears of bacterial endophthalmitis and control group comparison. For each box plot showing quantitative variations in the metabolic concentrations, the boxes denote interquartile ranges, the horizontal red line inside the box denotes the median, and the bottom and top boundaries of the boxes are the 25th and 75th percentiles, respectively. Lower and upper whiskers are the 5th and 95th percentiles, respectively. (A) Control vs. bacterial endophthalmitis: palmitoleamide (AUC, 0.88), myristoleoyl carnitine (AUC, 0.86), and methyl linolenate (AUC, 0.88) were the three metabolites with the highest AUCs per comparison represented in box plot and corresponding multivariate ROC curves were fitted by linear support vector machine algorithm. Control (*red bar graphs*) and bacterial endophthalmitis (*green bar graphs*). (B) Noninfectious uveitis vs. bacterial endophthalmitis: Uveitis yielding AUC values of 0.62 for palmitoleamide, 0.48 for myristoleoyl carnitine, and 0.71 for methyl linolenate with the highest AUCs per comparison represented in box plot and corresponding multivariate ROC curves were fitted by linear support vector machine algorithm. Uveitis (*green bar graphs*) and bacterial endophthalmitis (*red bar graphs*).

Discussion

Bacterial endophthalmitis is a major contributor to visual morbidity, particularly in tropical countries. Systemic injections of antibiotics and anti-inflammatory drugs are ineffective owing to the avascular nature of the vitreous and anterior chambers and the blood–ocular fluid barrier.²⁰ Direct infusion of drugs into intraocular space poses a risk of vitreous hemorrhage, retinal toxicity, corneal abrasions, central retinal artery occlusion, uveitis, or lens opacification.²¹^,²² Thus, there is an emergent need for early risk prediction, which can help to identify patients at greater risk of developing severe forms of the disease, allowing for earlier and more aggressive interventions to prevent further damage.²³^,²⁴ In this study, we identified a few potential endophthalmitis biomarkers in tears with high AUC values, including myristoleoyl carnitine, palmitoleamide, and methyl linolenate, when compared with normal controls. Tear collection is noninvasive and contains various biomarkers, including proteins, cytokines, and inflammatory markers, that can indicate the presence of inflammation or infection within the eye.²⁵^–²⁷ This study focuses on identifying metabolites that could serve as potential biomarkers for the early detection or monitoring of the condition, with a particular interest in noninvasive tear-based diagnostics. In our data, we found two metabolites (palmitoleamide and methyl linolenate) with lower expression in bacterial endophthalmitis patient tear samples when compared with normal controls. However, these metabolites were not expressed in noninfectious uveitis tear controls. Including noninfectious uveitis as a control in a metabolomics study of bacterial endophthalmitis is essential for distinguishing infection-specific metabolic changes from those owing to general inflammation. This approach enhances the accuracy, relevance, and clinical applicability of the study findings commonly identified in the tears of patients with bacterial endophthalmitis.

The top metabolites identified in aqueous humor, vitreous humor, and tears are different, but share common pathways, indicating that, although the specific metabolites present in each fluid may vary, they participate in similar biological processes or pathways. This correlation suggests a systemic or coordinated response within the eye, where different metabolites contribute to the same overall physiological functions or responses.

Methyl linolenate targets peroxisome proliferator-activated receptor (PPAR) gamma and retinoic acid receptor RXR-alpha.²⁸

PPARs are a group of nuclear receptors of steroid hormone receptor superfamily that acts as transcription factors to control the activation of genes.²⁹^,³⁰

Many studies demonstrated the role of PPAR in controlling inflammation, angiogenesis, and fibrosis. Recent studies have shown their roles in lipid metabolism and also in the control of inflammation, angiogenesis, and fibrosis.³¹^,³² The therapeutic potential of PPAR alpha and gamma agonists in treating corneal inflammation and neovascularization was investigated in a rat alkali burn model. The study found that the combination of these agonists significantly decreased inflammatory markers and inhibited neovascularization in the corneal tissue. Targeting PPAR pathways could be an effective strategy for managing inflammatory response in bacterial endophthalmitis. These findings suggest that targeting PPAR pathways could be an effective strategy for managing corneal injuries and related inflammatory conditions.

Palmitoleamide, another important biomarker, showed lower expression in bacterial endophthalmitis as compared with controls and was not detected in noninfectious uveitis. Palmitoleamide belongs to the oleamide group and is known to have an anti-inflammatory effect through P2Y2 and P2Y6 receptors. It was reported that P2Y2 and P2Y6 are the key receptors in mediating oleamide's anti-inflammatory effects in both murine microglia and human dendritic cells. Using gene expression analysis and receptor-specific antagonists and agonists, the researchers demonstrated that oleamide reduces inflammatory cytokine production by engaging these receptors. These findings suggest potential therapeutic targets for anti-inflammatory treatments based on oleamide's interaction with P2Y receptors.³³ It has been demonstrated that oleamide can activate PPAR gamma in a dose-dependent manner through direct binding, leading to increased expression of PPAR gamma target genes. This finding suggests that oleamide's physiological effects, including anti-inflammatory effects and metabolic regulation, may be mediated by PPAR gamma activation, indicating potential therapeutic applications in metabolic and inflammatory diseases.³⁴

We were interested in looking at the expression of myristoleoyl carnitine, because this metabolite was identified in vitreous humor and tears with a high VIP score. Higher levels of myristoleoyl carnitine in the vitreous humor and tears of patients with bacterial endophthalmitis as compared with controls may indicate a response to the presence of bacterial pathogens and their byproducts. However, levels were not significantly different when compared with noninfectious uveitis and so this was not a very good differentiator for infection-specific changes. Myristoleoyl carnitine belongs to the class of organic compounds known as acyl carnitines. Carnitine aids in beta-oxidation the transfer of long-chain fatty acids to the mitochondria.³⁵ Another study investigated the presence of carnitine in tears hypothesizing its role as an osmoprotectant.³⁶ Previous studies observed almost threefold increased plasma concentration of medium-chain acylcarnitines (e.g., C8-, C10-, C12, and C14-carnitine [myristoyl carnitine]), byproducts of incomplete fatty acid metabolism in T2DM patients suggesting that a mismatch in the fuel use and accumulation of acylcarnitines may portray the consequences of the disease state.³^,³⁷^,³⁸ Moreover, supplementation of L-carnitine in COVID-19 patients resulted in a decrease in inflammation and improved oxygen saturation levels caused owing to infection.³⁹

The inflammatory response is a natural defense mechanism aimed at containing and eliminating the infection. This process can exacerbate the bacterial infection and lead to significant damage to the eye's tissues and structures if not treated promptly and effectively. Without appropriate intervention, the infection can progress, leading to tissue destruction, visual impairment, and even loss of the eye. However, the severity of the infection and its potential to become fulminant (rapidly progressing and severe) depends on various factors, including the type and virulence of the bacteria, the individual's immune response, the timeliness of treatment, and the overall health of the eye. In some cases, certain bacteria may be particularly aggressive and cause rapid deterioration, leading to a fulminant infection. Our study suggests that methyl linolenate and palmitoleamide may be good detectors and differentiators in mild and severe endophthalmitis in tears. Further, molecular networks driving the imbalance in such host response factors may serve as therapeutic targets to reduce the severity of infection in individuals at greater risk of contracting a severe infection.

Limitations of the Study

Although the observations of this study are significant, there are a few limitations. Lower levels of methyl linolenate and palmitoleamide in the tears of those with severe bacterial infections were observed, and we do not know the exact bacterial species causing infection. Most of the pathogens were gram negative, as defined using Gram staining, but the exact bacterial species remains unknown, which may be very important to understand the host response in terms of metabolite differences. Second, the sample size of vitreous and aqueous humor is small; thus, additional studies, potentially involving multiple clinical centers, are needed to validate the results in a large number of samples.

Conclusions

We identified the tear metabolite markers methyl linolenate, palmitoleamide, and myristoleoyl carnitine, the levels of which were altered in bacterial endophthalmitis and had the potential to improve clinical outcomes for patients in early and accurate risk prediction of bacterial endophthalmitis, distinguishing it from other intraocular inflammatory conditions like uveitis. Timely diagnosis will allow timely treatment, reducing the risk of vision loss. By enabling earlier intervention and more precise treatments, metabolic markers can help to prevent complications such as retinal damage, persistent inflammation, and irreversible vision loss. With further validation, these metabolites could be incorporated into point-of-care diagnostic tools of risk prediction, thereby improving patient survival, preserving vision, and enhancing overall clinical outcomes in bacterial endophthalmitis. Further validation in larger cohorts is recommended to confirm their clinical usefulness.

Supplementary Material

Supplement 1

iovs-66-5-27_s001.tif^{(836.7KB, tif)}

Supplement 2

iovs-66-5-27_s002.tif^{(898.3KB, tif)}

Supplement 3

iovs-66-5-27_s003.tif^{(678.3KB, tif)}

Supplement 4

iovs-66-5-27_s004.xlsx^{(10.4KB, xlsx)}

Supplement 5

iovs-66-5-27_s005.xlsx^{(12.8KB, xlsx)}

Supplement 6

iovs-66-5-27_s006.xlsx^{(14.3KB, xlsx)}

Acknowledgments

Supported by the Narayana Nethralaya Foundation, Bangalore, India. The sponsor or funding organization had no role in the design or conduct of this research.

Author Contributions: Conceptualization, AG, NS and; sample collection PS, PD methodology and investigation, VD, SD formal analysis, RK, SD; data curation, RK, VD; writing original draft preparation, VD writing, review and editing, PD, AG, funding acquisition, RS, NS. All authors have read and agreed to the published version of the manuscript.

Disclosure: N. Shetty, None; P. Mahendradas, None; R. Kannan, None; S. Das, None; P. Sathe, None; R. Shetty, None; A. Ghosh, None; V. Deshpande, None

References

1. Mamalis N. Endophthalmitis. J Cataract Refract Surg. 2002; 28(5): 729–730. [DOI] [PubMed] [Google Scholar]
2. Romero CF, Rai MK, Lowder CY, Adal KA.. Endogenous endophthalmitis: case report and brief review. Am Fam Physician. 1999; 60(2): 510. [PubMed] [Google Scholar]
3. Jackson TL, Eykyn SJ, Graham EM, Stanford MR.. Endogenous bacterial endophthalmitis: a 17-year prospective series and review of 267 reported cases. Surv Ophthalmol. 2003; 48(4): 403–423. [DOI] [PubMed] [Google Scholar]
4. Rao NA, Hidayat AA.. Endogenous mycotic endophthalmitis: variations in clinical and histopathologic changes in candidiasis compared with aspergillosis. Am J Ophthalmol. 2001; 132(2): 244–251. [DOI] [PubMed] [Google Scholar]
5. Tanaka M, Kobayashi Y, Takebayashi H, Kiyokawa M, Qiu H.. Analysis of predisposing clinical and laboratory findings for the development of endogenous fungal endophthalmitis: a retrospective 12-year study of 79 eyes of 46 patients. Retina. 2001; 21(3): 203–209. [DOI] [PubMed] [Google Scholar]
6. Tiecco G, Laurenda D, Mulè A, et al.. Gram-negative endogenous endophthalmitis: a systematic review. Microorganisms. 2022; 11(1): 80. [DOI] [PMC free article] [PubMed] [Google Scholar]
7. Vitrectomy E. Results of the endophthalmitis vitrectomy study. Arch Ophthalmol. 1995; 113: 1479–1496. [PubMed] [Google Scholar]
8. Relhan N, Forster RK, Flynn HW Jr. Endophthalmitis: then and now. Am J Ophthalmol. 2018; 187: xx–xxvii. [DOI] [PMC free article] [PubMed] [Google Scholar]
9. Foster RE, Martinez JA, Murray TG, Rubsamen PE, Flynn HW Jr, Forster RK. Useful visual outcomes after treatment of Bacillus cereus endophthalmitis. Ophthalmology. 1996; 103(3): 390–397. [DOI] [PubMed] [Google Scholar]
10. Mathews AS, Pillai GS, Natasha R, Shetty M.. Endogenous endophthalmitis—a review. Ker J Ophthalmology. 2011; 23: 25–31. [Google Scholar]
11. Fujii GY, de Juan E Jr, Humayun MS, et al.. Initial experience using the transconjunctival sutureless vitrectomy system for vitreoretinal surgery. Ophthalmology. 2002; 109(10): 1814–1820. [DOI] [PubMed] [Google Scholar]
12. Lamaris G, Esmaeli B, Chamilos G, et al.. Fungal endophthalmitis in a tertiary care cancer center: a review of 23 cases. Eur J Clin Microbiol Infect Dis. 2008; 27: 343–347. [DOI] [PubMed] [Google Scholar]
13. Lingappan A, Wykoff CC, Albini TA, et al.. Endogenous fungal endophthalmitis: causative organisms, management strategies, and visual acuity outcomes. Am J Ophthalmol. 2012; 153(1): 162–166.e1. [DOI] [PubMed] [Google Scholar]
14. Shen X, Xu G.. Vitrectomy for endogenous fungal endophthalmitis. Ocul Immunol Inflamm. 2009; 17(3): 148–152. [DOI] [PubMed] [Google Scholar]
15. Sridhar J, Flynn HW, Kuriyan AE, Miller D, Albini T.. Endogenous fungal endophthalmitis: risk factors, clinical features, and treatment outcomes in mold and yeast infections. J Ophthalmic Inflamm Infect. 2013; 3: 1–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
16. Palexas GN, Green WR, Goldberg MF, Ding Y.. Diagnostic pars plana vitrectomy report of a 21-year retrospective study. Trans Am Ophthalm Soc. 1995; 93: 281. [PMC free article] [PubMed] [Google Scholar]
17. Chong J, Xia J.. MetaboAnalystR: an R package for flexible and reproducible analysis of metabolomics data. Bioinformatics. 2018; 34(24): 4313–4314. [DOI] [PMC free article] [PubMed] [Google Scholar]
18. Rutkowsky JM, Knotts TA, Ono-Moore KD, et al.. Acylcarnitines activate proinflammatory signaling pathways. Am J Physiol Endocrinol Metab. 2014; 306(12): 1378–1387. [DOI] [PMC free article] [PubMed] [Google Scholar]
19. Hanna VS, Hafez EAA.. Synopsis of arachidonic acid metabolism: a review. J Adv Res. 2018; 11: 23–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
20. Radhika M, Mithal K, Bawdekar A, et al.. Pharmacokinetics of intravitreal antibiotics in endophthalmitis. J Ophthal Inflamm Infect. 2014; 4: 22. [DOI] [PMC free article] [PubMed] [Google Scholar]
21. Moshfeghi AA, Rosenfeld PJ, Flynn HW Jr, et al.. Endophthalmitis after intravitreal anti–vascular endothelial growth factor antagonists: a six-year experience at a university referral center. Retina. 2011; 31(4): 662–668. [DOI] [PubMed] [Google Scholar]
22. Peng YJ, Kau YC, Wen CW, Liu KS, Liu SJ.. Solvent-free biodegradable scleral plugs providing sustained release of vancomycin, amikacin, and dexamethasone—an in vivo study. J Biomed Mater Res Part A. 2010; 94(2): 426–432. [DOI] [PubMed] [Google Scholar]
23. Nelson ML, Tennant MT, Sivalingam A, Regillo CD, Belmont JB, Martidis A.. Infectious and presumed noninfectious endophthalmitis after intravitreal triamcinolone acetonide injection. Retina. 2003; 23(5): 686–691. [DOI] [PubMed] [Google Scholar]
24. Miller FC, Coburn PS, Huzzatul MM, LaGrow AL, Livingston E, Callegan MC.. Targets of immunomodulation in bacterial endophthalmitis. Prog Retin Eye Res. 2019; 73: 100763. [DOI] [PMC free article] [PubMed] [Google Scholar]
25. Chng C-L, Seah LL, Yang M, et al.. Tear proteins calcium binding protein A4 (S100A4) and prolactin induced protein (PIP) are potential biomarkers for thyroid eye disease. Sci Rep. 2018; 8(1): 16936. [DOI] [PMC free article] [PubMed] [Google Scholar]
26. Amorim M, Martins B, Caramelo F, et al.. Putative biomarkers in tears for diabetic retinopathy diagnosis. Front Med. 2022; 9: 873483. [DOI] [PMC free article] [PubMed] [Google Scholar]
27. Angeles-Han S, Yeh S, Patel P, et al.. Discovery of tear biomarkers in children with chronic non-infectious anterior uveitis: a pilot study. J Ophthalmic Inflam Infect. 2018; 8(1): 17. [DOI] [PMC free article] [PubMed] [Google Scholar]
28. Lim E, Pon A, Djoumbou Y, et al.. T3DB: a comprehensively annotated database of common toxins and their targets. Nucleic Acids Res. 2010;(Database issue): D781–D786. [DOI] [PMC free article] [PubMed] [Google Scholar]
29. Abdelrahman M, Sivarajah A, Thiemermann C.. Beneficial effects of PPAR-gamma ligands in ischemia-reperfusion injury, inflammation and shock. Cardiovasc Res. 2005; 65: 772–781. [DOI] [PubMed] [Google Scholar]
30. Evans RM. The steroid and thyroid hormone receptor superfamily. Science. 1988; 240: 889–895 [DOI] [PMC free article] [PubMed] [Google Scholar]
31. Barbier O, Torra IP, Duguay Y, et al.. Pleiotropic actions of peroxisome proliferator-activated receptors in lipid metabolism and atherosclerosis. Arterioscler Thromb Vasc Biol. 2002; 22: 717–726. [DOI] [PubMed] [Google Scholar]
32. Zhang S, Gu H, Hu N. Role of peroxisome proliferator-activated receptor γ in ocular diseases. J Ophthalmol. 2015; 2015: 275435. [DOI] [PMC free article] [PubMed] [Google Scholar]
33. Kita M., Ano Y., Inoue A., et al.. Identification of P2Y receptors involved in oleamide-suppressing inflammatory responses in murine microglia and human dendritic cells. Sci Rep. 2019; 9: 3135. [DOI] [PMC free article] [PubMed] [Google Scholar]
34. Dionisi M, Alexander SP, Bennett AJ. Oleamide activates peroxisome proliferator-activated receptor gamma (PPARγ) in vitro. Lipids Health Dis. 2012; 11: 51. [DOI] [PMC free article] [PubMed] [Google Scholar]
35. Wajner M, Amaral AU.. Mitochondrial dysfunction in fatty acid oxidation disorders: insights from human and animal studies. Biosci Rep. 2016; 36(1): e00281. [DOI] [PMC free article] [PubMed] [Google Scholar]
36. Meadows JA, Wargo MJ.. Carnitine in bacterial physiology and metabolism. Microbiology. 2015; 161(Pt 6): 1161. [DOI] [PMC free article] [PubMed] [Google Scholar]
37. Pescosolido N, Imperatrice B, Koverech A, Messano M.. L-carnitine and short chain ester in tears from patients with dry eye. Optom Vis Sci. 2009; 86(2): E132–E138. [DOI] [PubMed] [Google Scholar]
38. Adams SH, Hoppel CL, Lok KH, et al.. Plasma acylcarnitine profiles suggest incomplete long-chain fatty acid β-oxidation and altered tricarboxylic acid cycle activity in type 2 diabetic African-American women. Nutr J. 2009; 139(6): 1073–1081. [DOI] [PMC free article] [PubMed] [Google Scholar]
39. Talebi SS, Ghasemi M, Etminani-Esfahani M, Mohammadi Y, Haddadi R.. Effects of L-carnitine supplementation in patients with mild-to-moderate COVID-19 disease: a pilot study. Pharmacol Rep. 2022; 74(6): 1296–1305. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement 1

iovs-66-5-27_s001.tif^{(836.7KB, tif)}

Supplement 2

iovs-66-5-27_s002.tif^{(898.3KB, tif)}

Supplement 3

iovs-66-5-27_s003.tif^{(678.3KB, tif)}

Supplement 4

iovs-66-5-27_s004.xlsx^{(10.4KB, xlsx)}

Supplement 5

iovs-66-5-27_s005.xlsx^{(12.8KB, xlsx)}

Supplement 6

iovs-66-5-27_s006.xlsx^{(14.3KB, xlsx)}

[bib1] 1. Mamalis N. Endophthalmitis. J Cataract Refract Surg. 2002; 28(5): 729–730. [DOI] [PubMed] [Google Scholar]

[bib2] 2. Romero CF, Rai MK, Lowder CY, Adal KA.. Endogenous endophthalmitis: case report and brief review. Am Fam Physician. 1999; 60(2): 510. [PubMed] [Google Scholar]

[bib3] 3. Jackson TL, Eykyn SJ, Graham EM, Stanford MR.. Endogenous bacterial endophthalmitis: a 17-year prospective series and review of 267 reported cases. Surv Ophthalmol. 2003; 48(4): 403–423. [DOI] [PubMed] [Google Scholar]

[bib4] 4. Rao NA, Hidayat AA.. Endogenous mycotic endophthalmitis: variations in clinical and histopathologic changes in candidiasis compared with aspergillosis. Am J Ophthalmol. 2001; 132(2): 244–251. [DOI] [PubMed] [Google Scholar]

[bib5] 5. Tanaka M, Kobayashi Y, Takebayashi H, Kiyokawa M, Qiu H.. Analysis of predisposing clinical and laboratory findings for the development of endogenous fungal endophthalmitis: a retrospective 12-year study of 79 eyes of 46 patients. Retina. 2001; 21(3): 203–209. [DOI] [PubMed] [Google Scholar]

[bib6] 6. Tiecco G, Laurenda D, Mulè A, et al.. Gram-negative endogenous endophthalmitis: a systematic review. Microorganisms. 2022; 11(1): 80. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib7] 7. Vitrectomy E. Results of the endophthalmitis vitrectomy study. Arch Ophthalmol. 1995; 113: 1479–1496. [PubMed] [Google Scholar]

[bib8] 8. Relhan N, Forster RK, Flynn HW Jr. Endophthalmitis: then and now. Am J Ophthalmol. 2018; 187: xx–xxvii. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib9] 9. Foster RE, Martinez JA, Murray TG, Rubsamen PE, Flynn HW Jr, Forster RK. Useful visual outcomes after treatment of Bacillus cereus endophthalmitis. Ophthalmology. 1996; 103(3): 390–397. [DOI] [PubMed] [Google Scholar]

[bib10] 10. Mathews AS, Pillai GS, Natasha R, Shetty M.. Endogenous endophthalmitis—a review. Ker J Ophthalmology. 2011; 23: 25–31. [Google Scholar]

[bib11] 11. Fujii GY, de Juan E Jr, Humayun MS, et al.. Initial experience using the transconjunctival sutureless vitrectomy system for vitreoretinal surgery. Ophthalmology. 2002; 109(10): 1814–1820. [DOI] [PubMed] [Google Scholar]

[bib12] 12. Lamaris G, Esmaeli B, Chamilos G, et al.. Fungal endophthalmitis in a tertiary care cancer center: a review of 23 cases. Eur J Clin Microbiol Infect Dis. 2008; 27: 343–347. [DOI] [PubMed] [Google Scholar]

[bib13] 13. Lingappan A, Wykoff CC, Albini TA, et al.. Endogenous fungal endophthalmitis: causative organisms, management strategies, and visual acuity outcomes. Am J Ophthalmol. 2012; 153(1): 162–166.e1. [DOI] [PubMed] [Google Scholar]

[bib14] 14. Shen X, Xu G.. Vitrectomy for endogenous fungal endophthalmitis. Ocul Immunol Inflamm. 2009; 17(3): 148–152. [DOI] [PubMed] [Google Scholar]

[bib15] 15. Sridhar J, Flynn HW, Kuriyan AE, Miller D, Albini T.. Endogenous fungal endophthalmitis: risk factors, clinical features, and treatment outcomes in mold and yeast infections. J Ophthalmic Inflamm Infect. 2013; 3: 1–6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib16] 16. Palexas GN, Green WR, Goldberg MF, Ding Y.. Diagnostic pars plana vitrectomy report of a 21-year retrospective study. Trans Am Ophthalm Soc. 1995; 93: 281. [PMC free article] [PubMed] [Google Scholar]

[bib17] 17. Chong J, Xia J.. MetaboAnalystR: an R package for flexible and reproducible analysis of metabolomics data. Bioinformatics. 2018; 34(24): 4313–4314. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib18] 18. Rutkowsky JM, Knotts TA, Ono-Moore KD, et al.. Acylcarnitines activate proinflammatory signaling pathways. Am J Physiol Endocrinol Metab. 2014; 306(12): 1378–1387. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib19] 19. Hanna VS, Hafez EAA.. Synopsis of arachidonic acid metabolism: a review. J Adv Res. 2018; 11: 23–32. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib20] 20. Radhika M, Mithal K, Bawdekar A, et al.. Pharmacokinetics of intravitreal antibiotics in endophthalmitis. J Ophthal Inflamm Infect. 2014; 4: 22. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib21] 21. Moshfeghi AA, Rosenfeld PJ, Flynn HW Jr, et al.. Endophthalmitis after intravitreal anti–vascular endothelial growth factor antagonists: a six-year experience at a university referral center. Retina. 2011; 31(4): 662–668. [DOI] [PubMed] [Google Scholar]

[bib22] 22. Peng YJ, Kau YC, Wen CW, Liu KS, Liu SJ.. Solvent-free biodegradable scleral plugs providing sustained release of vancomycin, amikacin, and dexamethasone—an in vivo study. J Biomed Mater Res Part A. 2010; 94(2): 426–432. [DOI] [PubMed] [Google Scholar]

[bib23] 23. Nelson ML, Tennant MT, Sivalingam A, Regillo CD, Belmont JB, Martidis A.. Infectious and presumed noninfectious endophthalmitis after intravitreal triamcinolone acetonide injection. Retina. 2003; 23(5): 686–691. [DOI] [PubMed] [Google Scholar]

[bib24] 24. Miller FC, Coburn PS, Huzzatul MM, LaGrow AL, Livingston E, Callegan MC.. Targets of immunomodulation in bacterial endophthalmitis. Prog Retin Eye Res. 2019; 73: 100763. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib25] 25. Chng C-L, Seah LL, Yang M, et al.. Tear proteins calcium binding protein A4 (S100A4) and prolactin induced protein (PIP) are potential biomarkers for thyroid eye disease. Sci Rep. 2018; 8(1): 16936. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib26] 26. Amorim M, Martins B, Caramelo F, et al.. Putative biomarkers in tears for diabetic retinopathy diagnosis. Front Med. 2022; 9: 873483. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib27] 27. Angeles-Han S, Yeh S, Patel P, et al.. Discovery of tear biomarkers in children with chronic non-infectious anterior uveitis: a pilot study. J Ophthalmic Inflam Infect. 2018; 8(1): 17. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib28] 28. Lim E, Pon A, Djoumbou Y, et al.. T3DB: a comprehensively annotated database of common toxins and their targets. Nucleic Acids Res. 2010;(Database issue): D781–D786. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib29] 29. Abdelrahman M, Sivarajah A, Thiemermann C.. Beneficial effects of PPAR-gamma ligands in ischemia-reperfusion injury, inflammation and shock. Cardiovasc Res. 2005; 65: 772–781. [DOI] [PubMed] [Google Scholar]

[bib30] 30. Evans RM. The steroid and thyroid hormone receptor superfamily. Science. 1988; 240: 889–895 [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib31] 31. Barbier O, Torra IP, Duguay Y, et al.. Pleiotropic actions of peroxisome proliferator-activated receptors in lipid metabolism and atherosclerosis. Arterioscler Thromb Vasc Biol. 2002; 22: 717–726. [DOI] [PubMed] [Google Scholar]

[bib32] 32. Zhang S, Gu H, Hu N. Role of peroxisome proliferator-activated receptor γ in ocular diseases. J Ophthalmol. 2015; 2015: 275435. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib33] 33. Kita M., Ano Y., Inoue A., et al.. Identification of P2Y receptors involved in oleamide-suppressing inflammatory responses in murine microglia and human dendritic cells. Sci Rep. 2019; 9: 3135. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib34] 34. Dionisi M, Alexander SP, Bennett AJ. Oleamide activates peroxisome proliferator-activated receptor gamma (PPARγ) in vitro. Lipids Health Dis. 2012; 11: 51. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib35] 35. Wajner M, Amaral AU.. Mitochondrial dysfunction in fatty acid oxidation disorders: insights from human and animal studies. Biosci Rep. 2016; 36(1): e00281. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib36] 36. Meadows JA, Wargo MJ.. Carnitine in bacterial physiology and metabolism. Microbiology. 2015; 161(Pt 6): 1161. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib37] 37. Pescosolido N, Imperatrice B, Koverech A, Messano M.. L-carnitine and short chain ester in tears from patients with dry eye. Optom Vis Sci. 2009; 86(2): E132–E138. [DOI] [PubMed] [Google Scholar]

[bib38] 38. Adams SH, Hoppel CL, Lok KH, et al.. Plasma acylcarnitine profiles suggest incomplete long-chain fatty acid β-oxidation and altered tricarboxylic acid cycle activity in type 2 diabetic African-American women. Nutr J. 2009; 139(6): 1073–1081. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib39] 39. Talebi SS, Ghasemi M, Etminani-Esfahani M, Mohammadi Y, Haddadi R.. Effects of L-carnitine supplementation in patients with mild-to-moderate COVID-19 disease: a pilot study. Pharmacol Rep. 2022; 74(6): 1296–1305. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Metabolomic Profiling of Ocular Fluid Identifies Host Response Factors for Risk Stratification in Bacterial Endophthalmitis

Naren Shetty

Padmamalini Mahendradas

Ramaraj Kannan

Samayitree Das

Priyanka Sathe

Rohit Shetty

Arkasubhra Ghosh

Vrushali Deshpande

Abstract

Purpose

Methods

Results

Conclusions

Materials and Methods

Patient Selection

Figure 1.

Table.

Sample Collection

Metabolomic Profiling

Sample Processing for Liquid Chromatography Coupled With a Tandem Mass Spectrometer

Results

Figure 2.

Altered Levels of Metabolites in Tears of Bacterial Endophthalmitis Patients

Figure 3.

Figure 4.

Metabolites Altered in the Aqueous Humor of Bacterial Endophthalmitis Patients

Figure 5.

Figure 6.

Metabolites Altered in Vitreous Humor of Bacterial Endophthalmitis Patients

Figure 7.

Figure 8.

Validation of the Classification Model in Bacterial Endophthalmitis Diagnosis

Figure 9.

Discussion

Limitations of the Study

Conclusions

Supplementary Material

Acknowledgments

References

Associated Data

Supplementary Materials

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases