Toxic oil syndrome (TOS) is a chronic, multisystem disease that emerged in Spain in May 1981 following the ingestion of rapeseed oil adulterated with aniline derivatives and illegally distributed for human consumption. The outbreak rapidly escalated, affecting over 20,000 individuals and causing more than 300 deaths within the first year (Abaitua Borda et al., 1998; Sánchez-Porro Valadés et al., 2003; Polentinos-Castro et al., 2021). The acute phase was characterized by eosinophilia, fever, myalgia, and pulmonary edema, later progressing to a chronic condition marked by scleroderma-like skin changes, peripheral neuropathy, muscular atrophy, and both pulmonary and neuropsychiatric manifestations (Gelpí et al., 2002; De La Paz et al., 2001). Although extensive multidisciplinary investigations were conducted, the specific toxic agent was never definitively identified. Nonetheless, fatty acid anilides and other byproducts of industrial oil denaturation were implicated based on chemical and toxicological analyses (Gelpí et al., 2002).

Kaufman and Krupp (1995) proposed that TOS and eosinophilia-myalgia syndrome may be part of a spectrum of chemically induced immune-mediated disorders involving the central nervous system. Their hypothesis initiated early discussions about mechanisms such as neuroinflammation and excitotoxicity. Notably, eosinophilia-myalgia syndrome itself is linked to neurocognitive dysfunction (Armstrong et al., 1997).

One of the earliest studies of cognitive impairment in TOS was reported in a small neuropsychological study by Del Ser et al. (1986), who found deficits in episodic and semantic memory within a few years of the onset. This was followed by a 12-year follow-up study by Kaufman et al. (1995), which offered clinical evidence of chronic morbidity in TOS survivors. Among 91 individuals re-evaluated more than a decade after exposure, over half reported persistent symptoms—fatigue, muscle cramps, arthralgias, subjective cognitive difficulties, and psychiatric complaints—despite being otherwise medically stable and lacking signs of a progressive neuromuscular disorder (Kaufman et al., 1995).

Further objective evidence came from a landmark case-referent study by De La Paz et al. (2003), conducted 18 years after the outbreak. TOS patients—particularly middle-aged women—performed significantly worse than matched controls on tests of motor strength, sensory function, and several neurocognitive domains such as memory, attention, processing speed, and psychomotor performance. The pattern of deficits indicated both peripheral and central nervous system involvement, with notable disruption of frontal-subcortical circuits (De La Paz et al., 2003).

Despite earlier findings, the long-term cognitive effects of TOS remain poorly understood. A recent 43-year follow-up case–control study found subtle yet measurable deficits in executive function, attention, and processing speed among TOS survivors, assessed with modern computerized tools (Lapeña-Motilva et al., 2025). In a separate biomarker study within the same group, slightly elevated serum neurofilament light chain levels were observed in TOS patients—mainly due to a few outliers—while levels of glial fibrillary acidic protein and phosphorylated tau 217 remained within normal ranges (Ruiz-Ortiz et al., 2025). These results suggest ongoing or residual neuronal injury in some individuals, without a biomarker pattern typical of progressive neurodegenerative disease (Ruiz-Ortiz et al., 2025).

Eye tracking provides an objective, noninvasive, high-resolution readout of cognitive control, as voluntary saccades are generated by frontal-subcortical circuits that subserve attention, executive control, and working memory. In particular, the frontal eye fields, dorsolateral prefrontal cortex, and anterior cingulate interact with basal ganglia loops and the superior colliculus to initiate or suppress gaze shifts; consequently, the antisaccade task—which requires suppression of a reflexive glance and generation of a willful movement in the opposite direction—is a sensitive probe of inhibitory control and working memory maintenance (Guerrero-Molina et al., 2021; García Cena et al., 2022a; García Cena et al., 2022b; Wolf et al., 2023; Benito-León et al., 2024; Leng et al., 2024; Benito-León et al., 2025). Across diverse neurological conditions, increased antisaccade error rates and prolonged latencies track frontal dysfunction, underscoring the utility of eye-movement metrics (e.g., error rates, reaction times) as surrogate markers of executive integrity (Guerrero-Molina et al., 2021; García Cena et al., 2022a; García Cena et al., 2022b; Wolf et al., 2023; Benito-León et al., 2024; Leng et al., 2024b; Benito-León et al., 2025).

To date, no studies have used eye tracking to examine cognitive dysfunction in TOS. We therefore employed three validated paradigms—visually guided saccades (basic sensorimotor function), memory-guided saccades (spatial working memory), and antisaccades (inhibitory control, executive function, and attention) to detect subtle deficits that may elude conventional testing (Guerrero-Molina et al., 2021; García Cena et al., 2022a; García Cena et al., 2022b; Wolf et al., 2023; Benito-León et al., 2024; Leng et al., 2024b; Benito-León et al., 2025). Given the chronic symptomatology of TOS, the predominance of dysexecutive features, and the absence of biomarker evidence for ongoing neurodegeneration, we hypothesized that TOS survivors would exhibit persistent oculomotor abnormalities, particularly in antisaccade performance, reflecting a long-standing disruption of frontal-subcortical circuits.

Of the 50 TOS patients and 50 healthy controls initially recruited, valid eye-tracking data were obtained for 47 and 44 participants, respectively. Participants were excluded due to acquisition-related issues (e.g., inadequate task compliance, calibration errors) or poor preprocessing quality (signal loss, artifacts, unstable tracking) that prevented analysis. As this was an exploratory, proof-of-concept study in a rare condition, a formal a priori sample-size calculation was not feasible. After data collection, a post-hoc power analysis (two-tailed independent-samples t-test, α = 0.05) showed that the analytic sample (N = 91) had approximately 83.8% power to detect the observed medium effect size (Cohen’s d ≈ 0.62) for the global cognitive score. Accordingly, the study was not powered to detect small between-group differences.

Demographic and neuropsychological data are summarized in Table 1. Groups were comparable in terms of age (mean age ~59 years), sex distribution (predominantly female), and educational attainment (p > 0.05 for all comparisons). However, TOS patients exhibited significantly higher prevalence of diabetes mellitus (25.5% vs. 2.3%), arterial hypertension (66.0% vs. 20.5%), and polyneuropathy (72.3% vs. 0%; all p < 0.005). TOS patients also showed markedly worse scores in health-related quality of life (EuroQol-5D index), fatigue (D-FIS), anxiety (BAI), and depression (BDI-II; all p < 0.001). Cognitively, they demonstrated significantly lower global cognitive scores (p = 0.007), with prominent deficits in executive function (p = 0.017), attention (p = 0.014), and information processing speed (p = 0.009). No significant differences were observed in memory subdomains.

This proof-of-concept study investigated saccadic eye movements and cognitive function in TOS survivors more than four decades after exposure. The most salient finding was a selective impairment in the antisaccade task: TOS participants generated significantly fewer correct antisaccades and exhibited increased reflexive and backward reflexive saccades. In contrast, performance in visually guided and memory-guided saccades was preserved across all parameters, including latency, gain, peak velocity, and spatial error measures. These findings suggest that basic oculomotor control and spatial working memory are preserved, whereas higher-order executive functions, particularly inhibitory control, are selectively impaired. This pattern is consistent with our neuropsychological results, which identified deficits in executive function, attention, and processing speed, with memory performance remaining relatively intact. SEM models confirmed that antisaccade impairments remained significant after adjusting for confounders. Notably, attention exerted an independent positive influence only on correct antisaccade performance (Figure 5), whereas reflexive and backward reflexive saccades (Figures 6 and 7) were driven by TOS status alone. Overall, the converging oculomotor and cognitive data indicate a long-term disruption of frontal-subcortical circuits in TOS survivors.

Our findings build upon earlier TOS studies by employing objective, high-temporal-resolution eye-tracking paradigms to detect subtle executive dysfunction. They also parallel deficits observed in other immune-mediated disorders. For instance, eosinophilia–myalgia syndrome—a condition with shared toxic and immunologic mechanisms—has been associated with similar impairments in executive function and attention, alongside magnetic resonance imaging evidence of subcortical white matter abnormalities (Armstrong et al., 1997). Both syndromes likely represent a continuum of toxic-immune encephalopathies with convergent effects on prefrontal networks (Armstrong et al., 1997).

Comparable profiles have been reported in chronic exposure to organophosphate pesticides, which selectively impair attention, executive functioning, and processing speed, while sparing language and visuospatial abilities (Jamal, 1997; Ross et al., 2013; London et al., 2012). These similarities reinforce the notion that diverse environmental toxicants can lead to common neurobehavioral outcomes via frontal-subcortical disconnection. Notably, organophosphate contamination was an early suspect in TOS etiology, although it was later discredited (Gelpí et al., 2002).

The antisaccade task is a sensitive probe of frontal dysfunction, particularly implicating the dorsolateral prefrontal cortex, frontal eye fields, and basal ganglia circuits (Pierrot-Deseilligny et al., 1995). Our finding of impaired antisaccades, alongside elevated reflexive and backward saccades, indicates a failure of inhibitory control, consistent with dysfunction in frontostriatal pathways (Pierrot-Deseilligny et al., 1995). This profile echoes those seen in neurodegenerative disorders, including Parkinson’s disease dementia and dementia with Lewy bodies (Mosimann et al., 2005), Huntington disease (Patel et al., 2012) and multiple system atrophy (Brooks et al., 2017), as well as psychiatric conditions such as schizophrenia (Caldani et al., 2017; Subramaniam et al., 2018; Obyedkov et al., 2019). Similar antisaccade impairments have also been reported in prodromal or cognitive disorders, notably idiopathic REM sleep behaviour disorder (Hanuška et al., 2019) and mild cognitive impairment (Koçoğlu et al., 2021). Eye_movement abnormalities have also been noted in Alzheimer’s disease dementia (Costanzo et al., 2023) and ataxias (Szpisjak et al., 2021), underscoring the broader relevance of antisaccade deficits marker of frontal-subcortical dysfunction across diverse disorders.

While reflexive saccades are generally preserved in early Alzheimer’s disease, they tend to be impaired in subcortical and mixed dementias (Mosimann et al., 2005). The altered antisaccade pattern in TOS thus further supports selective involvement of executive networks. Functional neuroimaging and lesion studies suggest that the dorsolateral prefrontal cortex is involved in response suppression, the frontal eye fields in vector inversion, and the anterior cingulate cortex in error monitoring (Funahashi, 2006; Munoz and Everling, 2004; McDowell et al., 2008; Curtis and D’Esposito, 2003).

TOS was originally linked to aniline-contaminated rapeseed oil, triggering a systemic eosinophilic and autoimmune response with central nervous system involvement (Abaitua Borda et al., 1998; Sánchez-Porro Valadés et al., 2003; Polentinos-Castro et al., 2021; Ricoy et al., 1983; Téllez et al., 1987). Neuropathological findings described central chromatolysis in anterior horn cells and brainstem nuclei, alterations in brainstem monoaminergic centers, and non-necrotizing vasculitis with focal ischemia—findings suggestive of microangiopathy and neuroinflammation. Although consistent white matter damage was not observed, the broader pattern suggests diffuse involvement rather than localized lesions (Ricoy et al., 1983; Téllez et al., 1987).

Persistent low-grade neuroinflammation may contribute to the long-term deficits observed in TOS. Animal models of environmental neurotoxicity, including exposure to manganese and organophosphates, consistently exhibit prolonged microglial activation, astrocytosis, and elevated cytokine release (e.g., IL-6, TNF-α), resulting in the disruption of frontal network function (Raison et al., 2006; Zhang et al., 2023; Kirkley et al., 2017; Popichak et al., 2018; Ke et al., 2019; Sarkar et al., 2018; Conley et al., 2020; Dinamene et al., 2013). In particular, manganese neurotoxicity has been shown to induce the secretion of pro-inflammatory cytokines (TNF-α, IL-1β, IL-6) by microglia, which in turn amplifies astrocytic reactivity via NF-κB signaling and results in sustained neuroinflammation and cortical dysfunction that persists well beyond the exposure period (Kirkley et al., 2017; Popichak et al., 2018; Ke et al., 2019; Sarkar et al., 2018; Conley et al., 2020; Dinamene et al., 2013). Comparable mechanisms have been described in organophosphate models, where sustained microglia-mediated neuroinflammation amplifies neuronal hyperactivation and disrupts cortical network function long after the acute exposure (Somkhit et al., 2022). In addition, endothelial dysfunction may further compromise cerebral perfusion in metabolically demanding prefrontal regions (Iadecola, 2013).

Importantly, the absence of elevated neurodegeneration biomarkers in TOS patients argues against progressive neuronal loss, distinguishing this syndrome from Alzheimer’s disease (Ruiz-Ortiz et al., 2025). Consistent with this, preserved episodic memory and the absence of hippocampal dysfunction suggest a static or slowly progressive encephalopathy, possibly more akin to conditions such as multiple sclerosis, small vessel disease or long COVID, where frontal-subcortical networks are preferentially affected (Benito-León et al., 2025; O’Brien et al., 2003; Jolly et al., 2025; Olazarán et al., 2009). Altogether, our findings support a model of non-degenerative, immune-mediated disruption of prefrontal white matter integrity in TOS, manifesting as enduring deficits in executive function and response inhibition.

Our study’s limitations include a small sample size, which may have reduced the ability to detect subtle differences in latency or gain metrics. This limited sample size increases the risk of a Type II error and, consistent with the exploratory nature of the study, justifies not applying corrections for multiple comparisons. The cross-sectional design also restricts causal conclusions. Polyneuropathy was not treated as a confounder; although characteristic of chronic toxic oil syndrome, it is not linked to cognitive dysfunction, and adjusting for it could lead to over-adjustment and mask the true effects of the disease. Nonetheless, we acknowledge its clinical implications and report its prevalence and distribution in Table 1. Comorbidities such as hypertension, diabetes, and higher levels of anxiety, depression, and fatigue were more frequent in the TOS group. Still, these factors were incorporated into our multivariable and SEM models alongside demographic variables. Importantly, the antisaccade deficit remained significant after full adjustment, suggesting that the observed association reflects disease-related effects rather than the influence of comorbid conditions alone. However, residual confounding cannot be completely ruled out. The lack of neuroimaging data further limits structural interpretation. Future research should include both structural and functional neuroimaging to better localize and understand the observed deficits. Finally, our findings may not apply beyond the specific group of Spanish TOS survivors studied. Comparative research in other populations exposed to toxins—such as workers exposed to organophosphates or patients with eosinophilia-myalgia syndrome—could help determine whether antisaccade deficits are a common feature of chronic toxic-immune encephalopathies.

In conclusion, to our knowledge, this is the first study to apply eye-tracking to evaluate cognitive dysfunction in TOS. By combining antisaccade and related paradigms with standardized cognitive testing, we demonstrate ongoing antisaccade impairments and dysexecutive deficits consistent with disruption of frontal-subcortical control. The pattern of findings is more compatible with immune-or vascular-mediated injury than with ongoing neurodegeneration. Given that eye-tracking is objective, noninvasive, brief, and well-tolerated in older, multimorbid cohorts, antisaccade-based measures offer a practical candidate biomarker of residual executive dysfunction that now warrants validation in longitudinal, adequately powered, multimodal studies.