Metabolites of traffic-related volatile organic compounds in age-related macular degeneration

Study design

This study was conducted as a prospective, non-randomized, cross-sectional investigation with consecutive case enrollment at Teikyo University Hospital and its affiliated institutions. The study adhered to the principles outlined in the Declaration of Helsinki. The study protocol was approved by the Ethics Committee of Teikyo University School of Medicine (Approval No. 18-228, approved on March 7, 2019). Several studies, including the present one, have been registered as clinical trials in the University Medical Information Network Clinical Trials Registry (UMIN-CTR) (Registration No.: UMIN000013684). Patient recruitment and study procedures were conducted between April 2020 and October 2021. Written informed consent was obtained from all participants prior to enrollment. Figure 1 presents a schematic overview of the experimental protocol.

Participants

This study enrolled 40 patients who were clinically diagnosed with AMD at Teikyo University and its affiliated hospitals. In cases where both eyes met the inclusion criteria, the eye with poorer visual acuity was selected for evaluation.

The diagnosis and classification of AMD and its subtypes were based on the diagnostic criteria established by the Japanese Retina and Vitreous Society Guidelines Committee for Neovascular Age-Related Macular Degeneration (2025). According to these criteria, participants were classified into the following four subtypes: drusenoid AMD (n = 3), typical AMD (tAMD) (n = 15), polypoidal choroidal vasculopathy (PCV) (n = 19), and retinal angiomatous proliferation (RAP) (n = 3).

Urinary metabolite analysis

Automotive gasoline contains aromatic VOCs such as benzene, ethylbenzene, toluene, and xylene (Geraldino et al., 2021). In addition, vehicle exhaust emissions include VOCs such as benzene, toluene, xylene, and styrene, which originate from pyrolyzed naphtha—a byproduct of ethylene and propylene manufacturing (Montero-Montoya, López-Vargas & Arellano-Aguilar, 2018). Once inhaled or absorbed into the human body, these VOCs are metabolized into compounds such as 2-methylhippuric acid, 3-methylhippuric acid, mandelic acid, phenylglyoxylic acid, and trans,trans-muconic acid, which are subsequently excreted in urine (Marchese et al., 2004) (see Fig. 2). In this study, these urinary metabolites were measured as biomarkers of internal VOC exposure.

Urinary VOC metabolite analysis was performed by US BioTek Laboratories (Seattle, WA, USA) using gas chromatography–mass spectrometry (GC/MS) according to their validated protocol (Salmi et al., 2010). Participants were instructed to abstain from food, beverages, medications, and dietary supplements on the morning of sample collection. First-morning urine samples were collected in disposable paper cups following initial voiding. A standardized collection strip (Dip ‘N Dry, US BioTek, Seattle, WA) was then immersed in each sample. The strips were sealed in foil pouches containing desiccant gel, stored at room temperature in a dry state, and subsequently shipped to US BioTek Laboratories for analysis.

Following the laboratory’s protocol, concentrations of 2-methylhippuric acid, 3-methylhippuric acid, mandelic acid, phenylglyoxylic acid, and trans-trans-muconic acid were measured using GC/MS. In addition, urinary creatinine levels were measured to normalize VOC metabolite concentrations, and all metabolite values were expressed as a ratio to creatinine. Because reference ranges for each metabolite vary by sex, the relative percentage increase or decrease from the sex-specific reference value was calculated for each participant.

All data analyses were conducted in a blinded manner. Investigators performing statistical evaluation were masked to participants’ personal identifiers and group allocation (e.g., disease status).

Analysis of gasoline composition and ambient VOC concentrations

To characterize the chemical composition of automobile fuel used in Japan, we reviewed publicly available product documentation from ENEOS Corporation, one of the country’s major petroleum companies. Specifically, we consulted the Safety Data Sheets (SDS) and technical reports available on the company’s official website (https://www.hd.eneos.co.jp/). The analysis included two commonly distributed fuel types—regular gasoline and premium high-octane gasoline. Information was extracted on the relative content of major aromatic hydrocarbons, including benzene, toluene, ethylbenzene, and xylene (collectively referred to as BTEX), which are known to be key components of VOCs. These data served as a chemical reference for understanding the typical VOC profile emitted by vehicle exhaust, one of the primary sources of urban air pollution. For components listed as concentration ranges in the SDS, representative values were used for interpretation, with due consideration of possible variations across commercial products.

In addition, to understand the background levels of ambient VOC exposure in the residential areas of study participants—specifically, the Itabashi and Nerima wards of Tokyo—we accessed air quality data from the Tokyo Metropolitan Government’s Environmental Monitoring Database (https://www.kankyo.metro.tokyo.lg.jp/). Daily VOC concentration measurements were retrieved from continuous monitoring stations in these districts for a three-year period (January 2019 to December 2021). Annual average concentrations were calculated for key VOCs, including BTEX compounds. All measurements were obtained using automated instruments at fixed monitoring sites and were based on officially published, publicly accessible data.

Statistical analysis

The sample size for this study was determined based on previous reports with comparable study designs (Miyama et al., 2015; Mimura, Noma & Mizota, 2019), which demonstrated statistically significant differences between groups of 10 participants each. Accordingly, we recruited a total of 20 control subjects—10 healthy individuals (Healthy group) and 10 cataract patients (Cataract group). Power analysis was conducted under the following assumptions: 10% margin of error, 95% confidence level, statistical power of 0.8, and a standard deviation of 5. The results indicated that a minimum of eight subjects per group would be sufficient to detect a significant difference between two groups. Therefore, the total of 20 control participants enrolled in this study was deemed statistically adequate.

Comparisons of mean values between two groups were performed using the two-tailed unpaired Student’s t-test. For comparisons involving three or more groups, the Kruskal–Wallis one-way analysis of variance by ranks was employed. Categorical variables were analyzed using either the chi-square test of independence or Fisher’s exact test, depending on expected cell frequencies. Correlations between continuous variables were assessed using Pearson’s correlation coefficient.

To identify independent factors associated with the presence of AMD, forward stepwise multiple logistic regression analysis was conducted. Explanatory variables included the urinary concentrations of five VOC metabolites: 2-methylhippuric acid, 3-methylhippuric acid, mandelic acid, phenylglyoxylic acid, and trans,trans-muconic acid.

All statistical analyses were performed using SAS software, version 9.1 (SAS Institute Inc., Cary, NC, USA). A p-value of less than 0.05 was considered statistically significant.

Patient characteristics

The demographic and clinical characteristics of the study population are presented in Table 1. There were no statistically significant differences among the Healthy, Cataract, and AMD groups with respect to age or sex. Similarly, the prevalence of hypertension (30–47%), type 2 diabetes mellitus (15–20%), and hyperlipidemia (15–20%) did not differ significantly among the groups. No participants in any group reported a history of cerebrovascular disease, coronary artery disease, renal disease, or hepatic disease.

Regarding smoking history, 30% of individuals in both the Healthy and Cataract groups and 22% in the AMD group reported a history of smoking. However, no statistically significant differences were observed among the three groups (p = 0.818, chi-square test). Furthermore, no participants in any group had a history of glaucoma or diabetic retinopathy.

Although we used the Kruskal–Wallis test, an appropriate non-parametric approach for comparing independent groups with small sample sizes, potential confounding factors such as age, sex, occupational exposure, smoking history, diabetes, hypertension, and renal function were not adjusted in the main analysis. However, no significant intergroup differences were observed in these variables, suggesting that their influence on urinary VOC metabolite levels was limited.

The Cataract group was included as an independent control group distinct from AMD, and subjects were enrolled only if no retinal pathology was observed. Although some participants had nuclear sclerosis above grade III, this was not expected to affect systemic VOC metabolism. Because cataract and AMD are caused by different etiological mechanisms, the use of the Cataract group as a comparator remains appropriate. Future studies with larger sample sizes and multivariable adjustment are warranted to validate these findings.

Composition of gasoline

Table 2 summarizes the concentrations (weight %) of major aromatic hydrocarbons contained in commercially available automotive gasoline in Japan. The average concentrations of individual components in regular and high-octane (premium) gasoline were as follows: benzene, 0.65% and 0.66%; ethylbenzene, 1.1% and 1.4%; xylene, 4.7% and 5.7%; and toluene, 9.0% and 23.0%, respectively. Notably, toluene was present at markedly higher levels in high-octane gasoline compared to other aromatic compounds. This likely reflects the intentional addition of toluene to increase octane ratings. These aromatic hydrocarbons are classified as VOCs, and their high volatility suggests a significant potential for atmospheric release, contributing to urban air pollution.

Ambient VOC concentrations

Table 3 shows the annual average concentrations of VOCs measured between January 2019 and December 2021 in two urban districts of Tokyo (Itabashi and Nerima). The yearly mean concentrations of representative aromatic VOCs ranged as follows: benzene, 0.66–0.83 µg/m³; trichloroethylene, 0.56–1.20 µg/m³; and toluene, 4.90–8.20 µg/m³. Although some variation was observed across years, no substantial year-to-year fluctuations were noted. Among the measured VOCs, toluene consistently showed the highest concentrations in both regions, likely due to emissions from vehicular exhaust and industrial activities. While benzene and trichloroethylene were detected at relatively lower levels, their potential health risks from long-term exposure should not be overlooked.

There were no significant differences in VOC concentrations between the two districts, suggesting a shared background of urban air pollution across the study population’s residential areas. These environmental measurements provide important contextual data supporting the relevance of VOC exposure in the lives of the enrolled patients and serve as a foundational basis for interpreting the clinical associations observed in this study.

Urinary VOC metabolite concentrations and percent increases relative to reference means across study groups

Figure 3 displays mean urinary concentrations (±SD) of VOC metabolites, normalized to creatinine, alongside their percent changes relative to established reference mean values in the Healthy, Cataract, and AMD groups.

• 2-Methylhippuric acid (Fig. 3A):

  • Healthy: 2.98 ± 0.83 µg/mg creatinine, Δ −0.1 ± 12.4%

  • Cataract: 3.31 ± 3.76 µg/mg, Δ 5.1 ± 63.7%

  • AMD: 5.99 ± 4.67 µg/mg, Δ 69.9 ± 75.7% The AMD group exhibited 201% and 181% higher mean concentrations compared to the Healthy and Cataract groups, respectively. Group differences were statistically significant (p = 0.035, Kruskal–Wallis). Post-hoc analyses confirmed significantly elevated levels in AMD versus Healthy (p < 0.001) and versus Cataract (p = 0.042) using two-tailed unpaired t-tests.

• 3-Methylhippuric acid (Fig. 3B):

  • Healthy: 0.17 ± 0.12 µg/mg, Δ 18.6 ± 52.0%

  • Cataract: 0.23 ± 0.22 µg/mg, Δ 36.9 ± 90.9%

  • AMD: 0.31 ± 0.68 µg/mg, Δ 84.1 ± 294.8% The AMD group showed 190% and 139% higher means relative to the Healthy and Cataract groups. However, differences across the three groups were not significant (p = 0.395, Kruskal–Wallis), nor were pairwise differences (AMD vs. Healthy, p = 0.105; AMD vs. Cataract, p = 0.254).

• Mandelic acid (Fig. 3C):

  • Healthy: 0.08 ± 0.06 µg/mg, Δ –22.4 ± 25.4%

  • Cataract: 0.12 ± 0.07 µg/mg, Δ −3.9 ± 29.8%

  • AMD: 0.24 ± 0.14 µg/mg, Δ 41.8 ± 57.6% The AMD group had concentrations 304% and 198% higher than Healthy and Cataract groups. Overall differences were significant (p = 0.035, Kruskal–Wallis), with AMD significantly higher than both comparison groups (p < 0.001).

• Phenylglyoxylic acid (Fig. 3D):

  • Healthy: 0.19 ± 0.07 µg/mg, Δ 0.5 ± 17.8%

  • Cataract: 0.25 ± 0.13 µg/mg, Δ 17.1 ± 34.9%

  • AMD: 0.23 ± 0.11 µg/mg, Δ 9.5 ± 30.0% Although mean ratios were 118% (vs. Healthy) and 90% (vs. Cataract), there were no significant differences among groups (p = 0.607, Kruskal–Wallis).

• trans, trans-muconic acid (Fig. 3E):

  • Healthy: 0.02 ± 0.02 µg/mg, Δ –10.1 ± 12.5%

  • Cataract: 0.06 ± 0.04 µg/mg, Δ −0.1 ± 28.9%

  • AMD: 0.05 ± 0.03 µg/mg, Δ –10.2 ± 22.9% Despite a 214% higher mean in AMD relative to Healthy, and 92% relative to Cataract, the difference across groups was significant (p = 0.025, Kruskal–Wallis). AMD was significantly higher than Healthy (p < 0.001) but not Cataract (p = 0.379).

Comparison of urinary VOC levels among AMD subtypes

Figure 4 presents urinary metabolite levels and their percent increases relative to reference means across AMD subtypes: drusenoid, typical (t-AMD), PCV, and RAP. Metabolites evaluated included 2-methylhippuric acid (Fig. 4A), 3-methylhippuric acid (Fig. 4B), Mandelic acid (Fig. 4C), Phenylglyoxylic acid (Fig. 4D), and trans,trans-muconic acid (Fig. 4E). No significant differences were found among subtypes for any metabolite (Kruskal–Wallis).

The proportion of AMD cases exceeding reference values (i.e., %Δ >0) for each metabolite was:

• 2-methylhippuric acid: 90.0% (36/40)

• 3-methylhippuric acid: 37.5% (15/40)

• Mandelic acid: 77.5% (31/40)

• Phenylglyoxylic acid: 57.5% (23/40)

• trans,trans-muconic acid: 32.5% (13/40)

Correlation among urinary VOC metabolites

Figure 5A displays pairwise Pearson correlation coefficients among the five derivates, and Fig. 5B visualizes them via a bubble plot. Significant positive correlations were identified in six metabolite pairs:

• 2-methylhippuric acid & mandelic acid (r = 0.44, p < 0.001)

• 2-methylhippuric acid & phenylglyoxylic acid (r = 0.32, p = 0.012)

• 3-methylhippuric acid & trans,trans-muconic acid (r = 0.29, p = 0.027)

• Mandelic acid & phenylglyoxylic acid (r = 0.56, p < 0.001)

• Mandelic acid & trans,trans-muconic acid (r = 0.38, p = 0.003)

• Phenylglyoxylic acid & trans,trans-muconic acid (r = 0.32, p = 0.012)

No significant correlations were observed between:

• 2-methylhippuric acid & 3-methylhippuric acid (r = 0.05, p = 0.681)

• 2-methylhippuric acid & trans,trans-muconic acid (r = 0.11, p = 0.387)

• 3-methylhippuric acid & mandelic acid (r = 0.11, p = 0.412)

• 3-methylhippuric acid & phenylglyoxylic acid (r = 0.07, p = 0.581)

These findings may reflect shared metabolic pathways or common exposure sources among certain VOCs.

Association of VOC metabolites with AMD

Pearson correlation and forward stepwise logistic regression analyses were used to assess associations between urinary VOC metabolites and AMD status. Results (Table 4) revealed significant positive correlations in the AMD group for:

• 2-methylhippuric acid (r = 0.29, p = 0.011)

• Mandelic acid (r = 0.47, p < 0.001)

• trans,trans-muconic acid (r = 0.27, p = 0.020)

No significant correlations were found for 3-methylhippuric acid (r = 0.10, p = 0.218) or Phenylglyoxylic acid (r = 0.08, p = 0.279).

The cumulative measure for all five VOCs also showed a significant correlation with AMD status (r = 0.31, p = 0.007).

In multivariate logistic regression, mandelic acid remained a significant positive predictor of AMD (odds ratio (OR) = 17.20, p < 0.001), whereas phenylglyoxylic acid was identified as a significant negative predictor (OR = 0.12, p = 0.028).

Summary of the study

This study aimed to investigate the association between VOC metabolites and AMD by comparing urinary concentrations of VOC metabolites among three groups: Healthy, Cataract, and AMD. The results revealed that urinary levels and percent increases relative to reference values for 2-methylhippuric acid, mandelic acid, and trans,trans-muconic acid were significantly elevated in the AMD group. These findings suggest a possible involvement of these compounds in the pathogenesis of AMD.

In particular, both 2-methylhippuric acid and mandelic acid were significantly higher in the AMD group compared to the Healthy and Cataract groups. They also showed significant positive correlations with AMD in Pearson’s analysis. Notably, mandelic acid emerged as a significant independent factor associated with AMD in multivariate regression analysis, suggesting its potential utility as a biomarker for AMD risk. Conversely, although phenylglyoxylic acid tended to be higher in the AMD group, the differences were not statistically significant, indicating a limited association with AMD. Several urinary VOC metabolites did not show statistically significant differences between groups, but these findings are still valuable in understanding the overall exposure profile. The multivariate analysis indicated that only mandelic acid was independently associated with AMD after adjustment for potential confounders, suggesting that styrene-related exposure may play a unique role in AMD pathogenesis. The coexistence of both significant and non-significant results reflects the complex multifactorial nature of AMD, in which environmental and metabolic influences may vary among individuals.

The gasoline commonly used in the residential areas of study participants was found to contain high concentrations of toluene, xylene, ethylbenzene, and benzene. Air quality monitoring data from two urban areas in Tokyo further confirmed the presence of VOCs in the environment, with toluene detected at the highest levels, followed by benzene and trichloroethylene. These compounds are known to be metabolized into the urinary biomarkers analyzed in this study: methylhippuric acids (from toluene and xylene), mandelic acid (from ethylbenzene), and trans,trans-muconic acid (from benzene).

These VOCs are well-documented components of urban air pollution, primarily originating from automobile exhaust and industrial emissions. Chronic inhalation exposure has been shown to induce oxidative stress by generating ROS and depleting intracellular glutathione, a key antioxidant (Wang et al., 2013). The significant elevation of these metabolites in the AMD group observed in this study suggests that long-term VOC exposure may contribute to oxidative damage of RPE cells and photoreceptors, thereby playing a role in the onset and progression of AMD.

Correlation analyses among the metabolites revealed particularly strong positive associations between 2-methylhippuric acid and mandelic acid, mandelic acid and phenylglyoxylic acid, and mandelic acid and trans,trans-muconic acid. These findings imply that certain VOC metabolites may share common metabolic pathways or originate from overlapping environmental exposure sources. In contrast, 3-methylhippuric acid exhibited substantial inter-individual variability and showed no significant differences among groups or clear association with AMD.

Importantly, no significant differences in age, sex, smoking history, or systemic comorbidities such as type 2 diabetes and hypertension were observed among the study groups. Therefore, the elevated urinary VOC metabolite levels in the AMD group are likely to reflect environmental VOC exposure rather than underlying lifestyle or medical conditions.

Proposed mechanisms linking VOC exposure to AMD progression

The elevated urinary levels of 2-methylhippuric acid, mandelic acid, and trans,trans-muconic acid observed in the AMD group suggest increased systemic exposure to aromatic VOCs such as toluene, ethylbenzene, and benzene. These compounds are commonly present in urban air due to traffic and industrial emissions, and chronic exposure to them has been linked to tissue oxidative injury and inflammatory responses in multiple organs. Considering the established metabolic origins of these urinary biomarkers, their elevation provides indirect but biologically meaningful evidence of environmental VOC exposure. This association supports the plausibility that long-term exposure to these airborne compounds contributes to retinal tissue damage through oxidative stress, complement-mediated inflammation, and ocular surface irritation.

Based on the findings of this study, several mechanistic pathways are proposed through which traffic-related VOCs—including benzene, toluene, styrene, and xylene—may contribute to the development and progression of AMD.

First, oxidative stress induced by VOC exposure may play a critical role in damaging RPE cells. VOCs are inhaled and metabolized in the body, during which ROS are generated. It has been reported that compounds such as toluene, benzene, and styrene activate the p38 MAPK signaling pathway, upregulating cyclooxygenase-2 (COX-2) expression and promoting the production of pro-inflammatory prostaglandins such as PGE₂ and PGF₂α, thereby inducing oxidative stress and inflammation (Mögel et al., 2011). Given the high oxygen demand and rich content of polyunsaturated fatty acids in the RPE, these cells are particularly vulnerable to oxidative damage. Such stress may trigger cellular injury, drusen formation, and the activation of pro-inflammatory signaling cascades (Ochoa Hernández et al., 2025). Although oxidative stress parameters were not directly evaluated in this study, previous reports have established oxidative stress as a key mechanism by which VOC exposure induces cellular and retinal injury (Potilinski et al., 2021; Chua et al., 2022). Therefore, the discussion of oxidative stress in the present study is intended to provide biological context rather than to imply a direct causal inference.

Second, chronic inflammation induced by long-term VOC exposure may contribute to AMD pathogenesis through the activation of inflammatory cytokines and the complement cascade. VOCs not only cause oxidative stress but are also known to modulate immune responses. Several experimental studies have shown that VOCs upregulate COX-2 and inflammatory cytokines in macrophages and epithelial cells, particularly in pulmonary and other mucosal tissues, thus acting as a trigger for chronic inflammation (Mögel et al., 2011). In AMD, chronic inflammation—mediated by complement activation and microglial recruitment—is recognized as a central pathogenic feature. Genetic polymorphisms, such as those in complement factor H (CFH), may exacerbate inflammation and accelerate disease progression (Ogbodo et al., 2022).

Third, VOCs may contribute to retinal damage indirectly through effects on the ocular surface. VOCs present in the environment can adhere to the cornea and conjunctiva, inducing local irritation, dryness, inflammation, and oxidative stress. Indeed, symptoms such as eye dryness and irritation, along with increased oxidative stress markers (e.g., elevated O-methylhippuric acid), have been reported in individuals with “sick building syndrome” linked to indoor VOC exposure (Kwon et al., 2018). Persistent exposure at the ocular surface may allow inflammatory or oxidative signals to propagate inward, potentially affecting deeper ocular tissues such as the retina.

Previous clinical trials, such as the Age-Related Eye Disease Study 2 (AREDS2), demonstrated that antioxidant and nutraceutical supplementation—particularly with lutein, zeaxanthin, and omega-3 fatty acids—can attenuate oxidative tissue damage and slow AMD progression (Chew et al., 2014). These findings suggest that modulation of oxidative stress through dietary or pharmacological measures could potentially mitigate retinal damage resulting from chronic VOC exposure.

Taken together, these findings suggest that VOCs may contribute to AMD progression through multiple interconnected mechanisms: (1) oxidative stress-mediated damage to RPE cells, (2) chronic inflammation via activation of inflammatory cytokines and the complement system, and (3) ocular surface irritation potentially influencing retinal health. In essence, VOC-induced oxidative stress and chronic inflammation may play a critical role in RPE and photoreceptor degeneration, contributing to the pathophysiology of AMD.

Future research should aim to identify specific VOC exposure sources, establish dose–response relationships, and integrate oxidative stress biomarkers with retinal imaging findings to clarify the causal link between VOC exposure and AMD development.

Limitations