Anxiety disorders are prevalent among the elderly, though reported prevalence rates vary considerably depending on the population studied and the diagnostic criteria employed. In community-dwelling elderly populations, systematic reviews and meta-analyses have reported prevalence estimates ranging from 3.2% to 14.2%, while rates are substantially higher in clinical and institutionalized settings, where estimates may exceed 20%–28% in populations with comorbid chronic medical conditions [1, 2]. While prior research has predominantly examined younger populations, evidence suggests that anxiety disorders in elderly individuals cause greater psychological and physical distress, severely impacting their social engagement and mental well-being [3]. Therefore, early identification and intervention for anxiety disorders have become critical in clinical management.

The relationship between cognitive impairment (CI) and anxiety disorders in elderly populations is complex and bidirectional [4, 5]. Neurobiologically, both conditions share common pathophysiological mechanisms, including dysfunction in prefrontal-limbic circuits, dysregulation of neurotransmitter systems (particularly serotonin, gamma-aminobutyric acid (GABA), and dopamine), and alterations in the hypothalamic-pituitary-adrenal axis [6, 7, 8, 9]. In elderly patients with CI, structural brain changes in regions such as the amygdala, hippocampus, and prefrontal cortex can precipitate anxiety symptoms, while chronic anxiety may accelerate cognitive decline through sustained stress responses and inflammatory cascades [10, 11].

The temporal relationship between CI and anxiety is particularly relevant in clinical practice. While anxiety can manifest as a reaction to cognitive decline awareness, it can also represent an early neuropsychiatric symptom of underlying neurodegenerative processes [12]. Studies have shown that anxiety disorders in patients with mild cognitive impairment increase the risk of progression to dementia, suggesting shared pathological mechanisms [13, 14]. Furthermore, the cognitive domains most affected by anxiety disorders—attention, executive function, and memory—overlap significantly with those impaired in CI, creating a synergistic effect that compounds functional disability [14]. Importantly, in the context of neurodegenerative diseases, anxiety symptoms may constitute behavioral and psychological symptoms of dementia (BPSD) rather than primary anxiety disorders, and distinguishing between these entities remains a significant clinical challenge [13, 14].

Current diagnostic approaches, which rely on medical history and psychological scales, are subjective and often lead to misdiagnosis or delayed detection. Elderly patients are frequently diagnosed only after the manifestation of severe symptoms [5]. This challenge is compounded by the fact that anxiety symptoms in elderly patients with CI may be attributed to cognitive decline itself, leading to under-recognition and under-treatment [15, 16]. Although extensive research has explored the neurophysiological mechanisms and risk factors associated with anxiety disorders, there is limited research on early screening using serum biomarkers [6]. Thus, the development of objective assessment methods to identify high-risk individuals early is imperative.

Recent advancements in biomarker detection technologies have highlighted the potential of serum markers in the diagnosis of neurological disorders [7, 10]. Tau protein (Tau) and $\beta$-amyloid (A$\beta$) are established indicators of CI and have been implicated in anxiety pathways through their effects on synaptic plasticity and neuroinflammation [17, 18]. Visinin-like protein 1 (VILIP-1) modulates neural signaling and is implicated in CI pathology and stress response mechanisms [19]. Inflammatory markers such as tumor necrosis factor-$\alpha$ (TNF-$\alpha$) and interleukin-6 (IL-6) not only contribute to neurodegeneration but also directly influence mood regulation through their effects on serotonin metabolism and hypothalamic-pituitary-adrenal (HPA) axis function [20, 21]. Oxidative stress markers, such as malondialdehyde (MDA), reflect cellular damage that affects both the cognitive and emotional processing regions of the brain [22]. MicroRNA-34c (MiR-34c), a key regulator, is associated with anxiety behaviors and modulates anxiety-related pathways by regulating serotonin receptors and stress-response genes [23].

This study sought to analyze clinical data and serum samples from elderly male CI patients to identify biomarkers associated with anxiety disorders, offering novel perspectives for the characterization and future investigation of early detection and intervention in elderly men.

Understanding anxiety symptoms in elderly patients with cognitive impairment requires recognizing that both conditions often reflect shared neurobiological processes rather than representing distinct entities [13, 14]. In the context of neurodegenerative changes affecting elderly individuals, anxiety symptoms frequently emerge as neuropsychiatric manifestations of the same pathological mechanisms that drive cognitive decline.

The prefrontal-limbic circuits governing both cognitive processing and emotional regulation show overlapping dysfunctions in CI and anxiety disorders [15, 17]. A related diagnostic consideration is that the HAMA, while supplemented by DSM-5–based clinical interviews in this study, was originally developed for populations without significant cognitive impairments. In elderly patients with CI, somatic symptoms common to both conditions, such as sleep disturbance, fatigue, and concentration difficulties, may overlap, complicating the distinction between primary anxiety and anxiety arising as a behavioral and BPSD [13]. Although our exclusion of patients with pre-existing psychiatric histories was designed to reduce this ambiguity, the cross-sectional design cannot establish a temporal sequence between anxiety and cognitive decline. This diagnostic uncertainty should be considered when interpreting the associations reported below. When neurodegenerative processes affect these integrated networks, symptoms can manifest across both cognitive and emotional domains, explaining why anxiety symptoms are particularly prevalent in elderly CI patients and why our study found that Group B exhibited more severe deficits in multiple cognitive domains.

Our findings revealed that patients with CI and comorbid anxiety demonstrated significantly worse performance across most MOCA domains, with attention and delayed recall demonstrating the strongest discriminative ability (AUC $>$0.74). These cognitive domains are particularly vulnerable because they depend on prefrontal-hippocampal circuits, which are highly susceptible to stress-induced dysfunction and neurodegeneration [17]. The associations between these cognitive measures and anxiety severity, even within individual groups, support a dimensional relationship in which cognitive performance and emotional regulation exist on interconnected continua rather than as separate functions.

An important interpretative consideration is the possibility of bidirectional measurement bias. Anxiety can impair attentional resources, working memory capacity, and executive function, potentially resulting in lower MOCA scores that do not solely reflect underlying neurodegenerative pathology. This phenomenon has been well documented in the cognitive psychology literature, where state anxiety has been shown to consume working memory resources and disrupt prefrontal cortical efficiency. Consequently, the observed association between lower cognitive scores and higher anxiety severity may be partially attributable to the acute effects of anxiety on test performance rather than exclusively reflecting shared neurodegeneration. This bidirectional relationship represents a fundamental limitation of cross-sectional assessments and can only be disentangled through longitudinal designs with repeated cognitive evaluations under varying anxiety states.

The biomarker findings provide potential mechanistic insights into the shared pathophysiology underlying CI-anxiety comorbidity, although these interpretations remain speculative given the cross-sectional design. The inflammatory hypothesis of neuropsychiatric disorders offers a framework for understanding these observations [19, 20]. Elevated IL-6 and TNF-$\alpha$ levels in the anxiety group reflect chronic inflammatory activation, which has been shown in prior studies to impair cognition through microglial activation, synaptic dysfunction, and interference with neurotransmitter metabolism [22, 23]. IL-6 specifically disrupts serotonin synthesis by increasing indoleamine 2,3-dioxygenase activity, creating a mechanistic link between inflammation and mood dysregulation. Our findings are consistent with prior reports linking elevated peripheral inflammatory markers to neuropsychiatric symptoms in elderly populations, including a meta-analysis by Custodero et al. [10] demonstrating associations between inflammatory biomarkers and cognitive impairment with neuropsychiatric features. However, although significant, the within-group correlations between inflammatory markers and anxiety severity were modest in magnitude, and the causal direction of this relationship could not be determined from cross-sectional data.

The significantly elevated Tau and A$\beta$ levels in patients with anxiety suggest an association with accelerated neurodegeneration processes. While traditionally associated with Alzheimer’s pathology, these proteins also reflect acute neuronal stress and injury [21]. However, the multivariate analysis revealed that the adjusted OR for Tau was relatively modest (1.15 per unit increase), and the discrepancy between Tau’s high univariate AUC (0.957) and its attenuated effect in the adjusted model suggests that much of Tau’s discriminative ability in univariate analysis may reflect confounding by age, comorbidities, or intercorrelation with other biomarkers rather than a direct and independent association with anxiety status. This finding underscores the importance of multivariate adjustment when interpreting biomarker associations in observational studies with baseline imbalance.

Oxidative stress represents another convergent pathway, as evidenced by markedly elevated MDA levels in the anxiety group. This finding reflects compromised cellular antioxidant defenses and increased lipid peroxidation [24, 25]. MDA retained a strong independent association with anxiety status in the multivariate model (OR = 3.45), suggesting that oxidative stress may represent a more robust biological correlate of anxiety in this population, although confirmatory longitudinal studies are needed.

VILIP-1 emerged as one of our strongest discriminators (AUC = 0.914), consistent with its role as a sensitive marker of neuronal calcium dysregulation and cellular stress [26, 27]. Its elevation suggests widespread neuronal vulnerability extending beyond traditional cognitive networks to encompass the emotional processing regions. Similarly, the significant reduction in MiR-34c levels in patients with anxiety may represent impaired stress resilience mechanisms and reduced neuroprotective capacity [28]. MiR-34c demonstrated the strongest independent association with anxiety in the multivariate model (OR = 0.12), suggesting that it may be a particularly informative molecular marker worthy of further investigation in prospective studies. Our findings are broadly consistent with emerging evidence implicating miRNA dysregulation in neuropsychiatric conditions, as reviewed by Albano et al. [28], although direct comparisons are limited by differences in study populations and biomarker panels.

Our multivariate analysis, incorporating the Bonferroni correction for multiple testing, identified independent biomarker factors after controlling for demographic and clinical confounders. The persistence of significant associations for the oxidative stress indicator MDA and regulatory microRNA MiR-34c suggests that these may represent informative biological correlates rather than secondary effects, although validation in independent cohorts is essential. It must be acknowledged, however, that the pronounced baseline imbalances between the study groups—particularly the nearly threefold difference in the proportion of patients aged $\ge$70 years and the more than twofold difference in diabetes prevalence—raise the possibility that multivariable regression may not fully resolve this degree of structural non-comparability. When groups differ so markedly across multiple confounding dimensions simultaneously, the functional form assumptions underpinning parametric adjustment become more tenuous, and residual confounding from both measured and unmeasured variables remains a plausible alternative explanation for the observed associations. Accordingly, all biomarker associations identified in this study should be interpreted as exploratory findings that require confirmation in studies employing either design-based solutions (such as frequency-matched enrollment or propensity score methods) or prospective longitudinal designs with more balanced comparison groups. Similarly, the ROC/AUC findings reported herein represent hypothesis-generating assessments of discrimination performance within this specific cohort and should not be interpreted as evidence of clinically actionable diagnostic or predictive utility in the absence of external validation.

Regarding the generalizability of these findings, the restriction to male participants requires careful consideration. Sex differences in anxiety disorders and cognitive impairment are well documented, encompassing hormonal influences (e.g., estrogen-mediated neuroprotection), differential inflammatory and oxidative stress profiles, and variations in hypothalamic-pituitary-adrenal axis reactivity (HPA). These factors may substantially modify the biomarker associations observed in the present study, and our results should not be extrapolated to the female population. Beyond the sex restriction, the derivation of our study population from a single military veteran rehabilitation facility introduces a further dimension of selection that merits explicit consideration. Retired military veterans may carry a distinctive burden of cumulative psychological stress—including potential combat-related trauma exposure—that could independently modulate inflammatory, oxidative stress, and neurodegeneration biomarker profiles. Moreover, this population may differ from community-dwelling elderly individuals in terms of healthcare access patterns, medication use, lifestyle factors, and the nature and severity of chronic medical comorbidities. These selection-related characteristics represent sources of potential bias that are not captured by the covariates included in our regression model, and they further limit the extent to which the present findings can be extrapolated to the broader elderly CI population. Multicenter studies incorporating both sexes and adequate power for sex-stratified analyses are a critical priority for future research.

Future investigations would also benefit from employing more comprehensive neuropsychological batteries beyond the MOCA screening tool. While the MOCA is well-validated and appropriate for the clinical screening context of this study, detailed assessments of episodic memory (e.g., the Rey Auditory Verbal Learning Test), executive function (e.g., the Trail Making Test), language (e.g., the Boston Naming Test), and processing speed would provide more granular cognitive profiling and potentially improve the specificity of cognitive biomarker assessment approaches.

The clinical implications include the potential future development of biomarker-based screening approaches for anxiety identification in elderly CI populations, although the present findings are preliminary and hypothesis-generating. If validated in prospective studies with external cohorts, the identification of specific pathways (inflammation, oxidative stress, and neurodegeneration) could provide targets for therapeutic approaches, such as anti-inflammatory strategies and neuroprotective interventions.

This cross-sectional study identified associations between CI and anxiety disorders, demonstrating that MOCA scores (attention, delayed recall) and serum biomarkers (Tau, MDA, VILIP-1, and MiR-34c) exhibit discriminative ability for anxiety status in elderly men. These findings are consistent with a neurobiological model of CI-anxiety comorbidity involving inflammatory, oxidative stress, and neuronal damage pathways. However, these findings should be regarded as hypothesis-generating rather than confirmatory findings. Importantly, these results derive from a single-center cohort of retired male military veterans with pronounced baseline group imbalances, substantially limiting generalizability to the broader elderly CI population and necessitating that all reported biomarker associations be interpreted as exploratory. Limited by its cross-sectional design, use of severity screening rather than comprehensive diagnostic criteria for anxiety, inability to distinguish primary anxiety from BPSD-related symptomatology, small sample size, pronounced baseline imbalances between groups, male-only single-center focus, and absence of external validation, further prospective, longitudinal multi-center studies with comprehensive psychiatric assessment and sex-inclusive cohorts are needed to validate these findings before any clinical application can be considered.

The data presented in this study are available on reasonable request from the corresponding author.

XC, TL, JWZ, WZ, and JYD contributed to the study conception and design. Material preparation, data collection, and analysis were performed by XC, WZ, JWZ, TL, and JYD. The first draft of the manuscript was written by XC. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

This study was approved by the Ethics Committee of Jiangsu Rongjun Hospital (Ethics No. 20250925). We confirm that informed consent was obtained from all participants or their legal guardians. This study was conducted in accordance with the Declaration of Helsinki.

Not applicable.

This research was funded by the Research Project of Wuxi Municipal Health Commission (Major Project, Grant No. Z202322) and the Research Project of Wuxi City Science and Technology Association (Soft Project, Grant Nos. KX-25-A12 and KX-24-B24).

The authors declare no conflict of interest.

Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.31083/RN48908.