Anxiety in late-life depression: Associations with brain volume, amyloid beta, white matter lesions, cognition, and functional ability

Abstract

Objectives:

Late-life depression (LLD) is common and frequently co-occurs with neurodegenerative diseases of aging. Little is known about how heterogeneity within LLD relates to factors typically associated with neurodegeneration. Varying levels of anxiety are one source of heterogeneity in LLD. We examined associations between anxiety symptom severity and factors associated with neurodegeneration, including regional brain volumes, amyloid beta (Aβ) deposition, white matter disease, cognitive dysfunction, and functional ability in LLD.

Participants and Measurements:

Older adults with major depression (N = 121, Ages 65–91) were evaluated for anxiety severity and the following: brain volume (orbitofrontal cortex [OFC], insula), cortical Aβ standardized uptake value ratio (SUVR), white matter hyperintensity (WMH) volume, global cognition, and functional ability. Separate linear regression analyses adjusting for age, sex, and concurrent depression severity were conducted to examine associations between anxiety and each of these factors. A global regression analysis was then conducted to examine the relative associations of these variables with anxiety severity.

Results:

Greater anxiety severity was associated with lower OFC volume (β = −68.25, t = −2.18, p = .031) and greater cognitive dysfunction (β = 0.23, t = 2.46, p = .016). Anxiety severity was not associated with insula volume, Aβ SUVR, WMH, or functional ability. When examining the relative associations of cognitive functioning and OFC volume with anxiety in a global model, cognitive dysfunction (β = 0.24, t = 2.62, p = .010), but not OFC volume, remained significantly associated with anxiety.

Conclusions:

Among multiple factors typically associated with neurodegeneration, cognitive dysfunction stands out as a key factor associated with anxiety severity in LLD which has implications for cognitive and psychiatric interventions.

Introduction

Major depressive disorder in older adults, or late-life depression (LLD), is common, associated with functional decline in old age (Anstey et al., 2007), and frequently co-occurs with neurodegenerative diseases of aging (Diniz et al., 2013). Little is known about how heterogeneity within LLD relates to features associated with neurodegeneration. Anxiety is highly prevalent among individuals with LLD and is one source of heterogeneity in this population. Up to 48% of individuals with LLD present with a concurrent anxiety disorder (Beekman et al., 2000; Jeste et al., 2006; Lenze et al., 2000), which has been associated with poorer depression treatment outcomes and higher rates of recurrence (Andreescu et al., 2007; Steffens & McQuoid, 2005). While LLD often coincides with features commonly associated with neurodegeneration, little is known about how anxiety within LLD relates to such features. Examining how comorbid anxiety relates to factors such as brain structure, amyloid beta (Aβ) deposition, white matter lesions, cognition, and functional ability may elucidate etiological mechanisms implicated in anxiety symptoms in LLD, thereby help parse the heterogeneity of LLD, and may eventually inform precision medicine approaches for both mood symptoms and cognitive dysfunction in LLD.

Anxiety has been linked to neurodegeneration and neuropathology in older adults, including reduced gray matter volume (Karim et al., 2021; Mah et al., 2015) and greater Aβ deposition (Krell-Roesch et al., 2018) and white matter lesion burden (Vogels et al., 2007). These studies suggest that neural circuit disruptions may be elicited by structural abnormalities that diminish emotion regulation capacity and contribute to increased anxious affect. For example, structural and functional abnormalities in the orbitofrontal cortex (OFC), an area involved in emotional processing and behavioral flexibility (Schoenbaum et al., 2009), have been linked to anxiety disorders (Milad & Rauch, 2007) including in older adults with generalized anxiety disorder (Andreescu et al., 2017). In LLD, the association of anxiety with neurodegenerative features is understudied but has been most consistently associated with reduced gray matter in the OFC and insula (Laird et al., 2019), another region involved in emotion processing that is thought to be linked to anxiety-related impairments including threat sensitivity (Klumpp et al., 2012). Both of these regions have been implicated as factors contributing to anxious symptomatology in older adults with a history of depression (Potvin et al., 2015) and in community-dwelling older adults (Pink et al., 2017). Reduced gray matter in the amygdala, anterior cingulate cortex (ACC), and temporal cortex as well as other distributed frontal and parietal areas have also been associated with anxiety in older adults with a history of or current depression, but less consistently (Krause-Sorio et al., 2023; Laird et al., 2019; Potvin et al., 2015). With only two studies conducted specifically in LLD (Krause-Sorio et al., 2023; Laird et al., 2019), more work is needed to elucidate the associations between anxiety and gray matter integrity in this population given the high comorbidity of anxiety and documented vulnerability to lower cortico-limbic volume in LLD (Andreescu et al., 2008).

Previous work has also linked anxiety with neuropathology related to two of the most common neurodegenerative diseases of aging, Alzheimer’s disease (AD), and cerebrovascular disease (CVD). AD is characterized by increased Aβ. Associations between Aβ and anxiety have been documented in non-depressed cognitively normal older adults (Johansson et al., 2022; Krell-Roesch et al., 2018; Lewis et al., 2022) and in a mixed sample of older adults with subjective cognitive decline, mild cognitive impairment, and AD dementia (Banning et al., 2020). Research examining associations between anxiety and white matter lesion burden characteristic of CVD has been more limited (Clancy et al., 2021), yet higher white matter lesion burden has been associated with greater anxiety in older adults with a history of ischemic stroke (Kim et al., 2011) and heart failure (Vogels et al., 2007). Overall, although Aβ and white matter pathology appear implicated in anxious symptomatology in older adults, possibly due to related disruptions in emotion circuitry, to our knowledge these factors have not yet been examined in association with anxiety in LLD.

Poorer cognitive functioning in older adults is also known to be associated with anxiety symptoms. For example, in community-based studies, anxiety in older adults has been related to increased likelihood of cognitive dysfunction and impairment (Beaudreau & O’Hara, 2008; Diefenbach et al., 2014; Smith et al., 2021), even when controlling for depression severity and other demographic and clinical covariates (Bierman et al., 2005). Furthermore, community-dwelling older adults with coexisting depression and anxiety exhibited more wide-ranging cognitive deficits compared to individuals with anxiety or depression only (Beaudreau & O’Hara, 2009). In LLD, previous work found no cross-sectional associations between increased anxiety and poorer cognition (DeLuca et al., 2005); however, analyses did not control for depression severity and findings may have been limited by a small sample size. Furthermore, to our knowledge no previous studies have evaluated the relative associations of anxiety in LLD with cognition alongside neurodegenerative features.

Previous work has also shown consistent associations between anxiety and declines in overall functional status among older adults. For example, associations between higher anxiety and poorer functional status have been found in older adults in nursing care settings (Schultz et al., 2004) and in the community (Krell-Roesch et al., 2020). Furthermore, lower functional ability has also been reported among older adults with anxiety disorders (Porensky et al., 2009). Although evidence suggests that comorbid depression and anxiety are associated with poorer functional status in community samples (Simning & Seplaki, 2020), these associations are understudied within LLD. Some studies suggest that elevated anxiety levels may relate to added functional impairment in LLD (Morin et al., 2020; Schoevers et al., 2003), but no studies have examined relative associations of functional ability and neurodegenerative features with anxiety, despite established links between Aβ and WMH and functional decline.

To elucidate the relation between anxiety and neurodegeneration within LLD, we evaluated the associations of anxiety symptom severity with cortico-limbic volumes, cortical Aβ deposition, WMH, global cognitive functioning, and functional ability in a sample of older adults with LLD. As anxiety and depression have overlapping symptomatology, we controlled for depression severity in our analyses, to highlight the independent effects of anxiety. Our a priori hypotheses were that greater anxiety symptom severity would be associated with smaller cortico-limbic volume in the OFC and insula, greater cortical Aβ deposition and white matter lesion burden, as well as poorer performance on measures of global cognition and functional ability. Additionally, we evaluated the relative associations of these variables with anxiety severity.

Materials & methods

Participants and procedures

Participants were older adults with Major Depressive Disorder (MDD; N = 121, Ages 65–91). MDD was determined via multidisciplinary consensus based on the 17-item Hamilton Depression Rating Scale (Hamilton, 1960; total ≥ 15), and the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders-IV (American Psychiatric Association, 2000). Participants were eligible if they had unipolar MDD without psychotic features and were currently experiencing an episode of at least six-week duration. Exclusion criteria included other concurrent primary Axis I disorders (except for certain anxiety disorders, e.g., generalized anxiety disorder), neurologic disorders that affect the CNS (e.g., epilepsy, Parkinson’s disease, traumatic brain injury), and a clinical dementia diagnosis or evidence of dementia (<25 on the Mini-Mental Status Exam and Clinical Dementia Rating Scale > 1.0).

Participants completed MRI imaging and AV-45 (Florbetapir) amyloid PET imaging and had their blood drawn for DNA/RNA banking with APOE genotyping. They also underwent cognitive testing and comprehensive psychiatric evaluations. The study was designed and executed in accordance with the Declaration of Helsinki for the protection of Human Subjects and approved by the Institutional Review Boards of each study site (University of California, San Francisco, University of Pittsburgh). Written informed consent was obtained from all participants.

Measures

Anxiety

The Generalized Anxiety Disorder 7-item scale (Spitzer et al., 2006; GAD-7), which has been validated in older adults (Wild et al., 2014), evaluated anxiety symptom severity. Participants rated each item on a four-point scale (0 = “not at all,” to 3 = “nearly every day”) to indicate experienced severity of anxiety symptoms over the past two weeks. Anxiety severity was used as a continuous variable across analyses.

Depression

The Geriatric Depression Scale (Yesavage, 1988; GDS) measured depression severity and was used as a covariate across analyses.

Cortico-limbic volumes and white matter hyperintensities

MRI data were obtained using a 3T scanner to measure cortico-limbic volumes and white matter hyperintensities. MRI protocols included 3D MP-RAGE or IR-SPGR T1-weighted MRI with sagittal slices and voxel size of 1 × 1 × 1.2 mm³ and 2D FLAIR MRI with axial slices and voxel size of 0.9 × 0.9 × 5 mm³ (Gunter et al., 2017). Cortico-limbic volumes were measured by parcellating the entire cortex of each subject into 41 bilateral regions-of-interest (ROIs) based on a previously published volumetric Desikan-Killiany-Tourville atlas via FreeSurfer v5.3 processing. A Bayesian inference approach was used to noise-correct and segment FLAIR images (Schwarz et al., 2009) for white matter hyperintensities. The OFC and the insula were identified a priori as primary region of interest. In line with previous work, and because sensitivity analyses revealed no significant laterality or subarea effect in the present sample, we used bilateral volumes and combined lateral and medial OFC volumes into single bilateral volumes. Supplemental analyses examined additional regions that have less consistently been shown to be implicated in older adults with anxiety (see Supplemental Materials; Andreescu et al., 2017; Potvin et al., 2015). White matter lesion burden was defined as the total volume of cerebral white matter hyperintensities (WMH).

Amyloid beta (Aβ)

As previously described (Mackin et al., 2021), all participants underwent Florbetapir PET imaging with four 5-minute frames acquired 50–70 minutes post-injection which were coregistered to a concurrently acquired structural MRI scan. Regional means defined in FreeSurfer 5.3 were used to calculate Florbetapir cortical summary SUVR, which was a composite of frontal, cingulate, temporal, and parietal regions relative to the whole cerebellum (Landau et al., 2013). The composite was used as a continuous variable in our analyses.

APOE genotype

Blood-based DNA/RNA banking determined APOE genotype. A dichotomous variable indicated the absence or presence of the e4 allele (i.e., risk genotypes included 3|4 or 4|4).

Global cognitive functioning

Alzheimer’s Disease Assessment Scale-Cognitive Subscale (Rosen et al., 1984; ADAS-Cog) was used to assess global cognitive functioning via 13 subtests designed to measure cognitive dysfunction characteristic of AD. Higher scores indicate more significant impairment.

Functional ability

The Clinical Dementia Rating Scale ([Morris, 1993]CDR) Sum of Boxes (SB) score was used to measure functional ability assessed via semi-structured interviews administered by trained and certified interviewers, with higher scores indicating greater impairment.

Statistical analyses

Anxiety severity in the sample was determined via symptom endorsement on the GAD-7 according to validated published ranges (Spitzer et al., 2006): minimal (≤4 points), mild (5–9 points), moderate (10–14 points), and severe anxiety (≥15 points). Additionally, the number of individuals with concurrent or lifetime history of a secondary anxiety diagnosis was identified via the SCID.

To examine the associations between anxiety and brain structure (OFC and insula volumes), neuropathology (Aβ SUVR, WMH), and measures of global cognition (ADAS-Cog), and functional ability (CDR-SB), linear regression analyses were conducted with anxiety symptom severity, measured continuously, as the predictor of interest of each of the other variables in separate models. Age, gender, and concurrent depression severity were added as covariates across all models. Intracranial volume was added as an additional covariate in models of cortico-limbic volumes, APOE genotype in the model including Aβ, and education level in the models predicting ADAS-Cog and CDR-SB. This was done to control for known associations between these covariates and respective outcome variables. Given our a priori hypotheses based on prior literature independently implicating the OFC and the insula (Laird et al., 2019; Potvin et al., 2015), cortico-limbic volume analyses were not corrected for multiple comparisons. Log-transformed variables of Aβ SUVR, WMH, and CDR-SB were used in regression analyses as the raw distribution of these variables was significantly skewed. Regression residuals were evaluated for normality and influential cases. Residuals were not normally distributed and/or influential cases were identified in analyses examining Aβ SUVR, WMH, and CDR-SB, and robust regression was conducted as a follow-up analysis for these outcomes. Non-parametric Mann-Whitney tests were additionally conducted to examine whether Aβ SUVR, WMH, and CDR-SB differed by anxiety (i.e., minimal/mild anxiety vs. moderate/severe anxiety).

To evaluate the relative associations of the factors associated with anxiety, a global regression analysis was conducted using anxiety symptom severity, measured continuously, as the outcome. Predictors in this model included factors that emerged as significantly associated with anxiety in initial analyses, as well as demographic characteristics (age, gender, education level) and concurrent depression. Volumes added to this model were adjusted for intracranial volume by creating a variable that represents the proportion of cortico-limbic volume to intracranial volume. Regression residuals were evaluated for normality and influential cases to ensure regression assumptions were met.

Results

Prevalence of anxiety

In the present LLD sample, 50% of individuals endorsed moderate or severe anxiety symptoms (GAD-7), and 27% met criteria for a concurrent anxiety disorder with 7% meeting criteria for more than one anxiety disorder (Table 1). An additional 3% met criteria for lifetime, but not current anxiety disorders. Correlations among variables of interest in this study are presented in Supplemental Table 1.

Table 1. Demographic and clinical characteristics (N = 121)

Amyloid beta positivity defined as SUVR > 1.11; GAD-7 = Generalized Anxiety Disorder 7-item scale, GDS = Geriatric Depression Scale (15 items), ADAS-Cog = Alzheimer’s Disease Assessment Scale-Cognitive Subscale, CDR-SB = Clinical Dementia Rating scale Sum of Boxes; Missing data.

Race, 1; GDS, 1; ADAS-Cog, 1; CDR-SB, 3; Concurrent Anxiety Disorder, 6.

^{^}8 individuals met criteria for more than one concurrent anxiety disorder; qualitative labels of the GAD-7 are provided here to illustrate anxiety score distribution in the present sample; however, the continuous GAD-7 total score was used across all analyses.

Association of anxiety with cortico-limbic volumes

Controlling for intracranial volume, age, gender, and concurrent depression severity, anxiety was inversely associated with OFC volume (Table 2; b = −68.25, t = −2.18, p = 0.031, η_p ² = 0.04, Figure 1a) but was not associated with insula volume (b = −19.76, t = −0.85, p = 0.395, η_p ² = 0.01, Figure 1b). Anxiety was not significantly associated with any of the volumes examined in supplemental analyses (i.e., ACC, amygdala, temporal cortex; see Supplemental Materials).

Figure 1. Associations of anxiety with insula (a) and orbitofrontal cortex (b) volumes.

Table 2. Summary of demographic and clinical predictors of orbitofrontal cortex volume and cognitive impairment

OFC, orbitofrontal cortex; ADAS-Cog, Alzheimer’s Disease Assessment Scale-Cognitive Subscale; DF, degrees of freedom; SE, standard error; ICV, intracranial volume.

*p < .05.

**p < .01.

Association of anxiety with Aβ deposition and white matter lesion burden

After controlling for age, gender, and concurrent depression severity, no significant associations emerged between anxiety severity and Aβ SUVR (b = 0.002, t = 1.15, p = 0.253, η_p ² < .01) or WMH (b = −0.01, t = −0.45, p = 0.656, η_p ² < .01). Follow-up Mann-Whitney U tests showed no difference between minimal/mild and moderate/severe anxiety for Aβ SUVR (U = 1720.50, p = .570) or WMH (U = 1776.50, p = .902).

Association of anxiety with global cognitive functioning

Higher anxiety severity scores were significantly associated with worse performance on ADAS-Cog (Table 2; b = 0.23, t = 2.46, p = 0.016, η_p ² = .05), controlling for age, gender, education level, and concurrent depression severity. Older age also emerged as significantly associated with worse ADAS-Cog performance in this model. Table 2 presents result metrics of all model parameters. Examination of model residuals identified two influential cases. Sensitivity analyses that removed these cases from the regression did not change the results or interpretation of the regression.

Association of anxiety with functional ability

Adjusting for age, gender, education, and concurrent depression severity, anxiety severity was not associated with CDR-SB (b = 0.02, t = 1.76, p = 0.081, η_p ² = .03). Follow-up Mann-Whitney U showed no difference between minimal/mild and moderate/severe anxiety (U = 1970.50, p = .197).

Regression analysis evaluating relative associations of cognitive functioning and brain structure with anxiety severity

ADAS-Cog and OFC volume (adjusted for intracranial volume) were entered into the model predicting anxiety severity given their significant associations with anxiety severity in previous models. Worse ADAS-Cog performance was significantly associated with higher anxiety severity (b = 0.24, t = 2.62, p = 0.010, η_p ² = .06), but OFC volume did not show a significant association (b = −399.61, t = −1.13, p = 0.262, η_p ² = .01). Table 3 displays result metrics of all model parameters and Figure 2 presents a visual representation of the significant effect.

Table 3. Regression analysis predicting anxiety severity in LLD

OFC, orbitofrontal cortex.

^{^}adjusted for intracranial volume; bolded variable is a significant predictor of anxiety severity in this model.

Discussion

The key findings of this study were that greater anxiety severity was associated with greater cognitive dysfunction and lower OFC volume. Additionally, we found that when considering their relative associations with anxiety, only global cognitive functioning remained significantly associated with anxiety severity, independent of OFC volume and concurrent depression severity. We found no associations between anxiety severity and cortical Aβ deposition, WMH, or functional impairment in the present sample. Taken together these findings suggest that cognitive dysfunction and smaller OFC volume, rather than neuropathological indicators more commonly associated with emerging AD or CVD, may be key to understanding anxiety symptoms in LLD.

Figure 2. Partial regression plot of the association between anxiety severity and cognitive functioning, controlling for age, gender, level of education, concurrent depression severity, and orbitofrontal volume.

One major finding is that higher anxiety severity in LLD was associated with poorer performance on a measure of global cognition. This aligns with studies of community-dwelling older adults (Beaudreau & O’Hara, 2008; Bierman et al., 2005; Diefenbach et al., 2014; Smith et al., 2021) and prior work showing that coexisting depression and anxiety among older adults relates to more wide-ranging cognitive deficits compared to individuals with increased anxiety or depression only (Beaudreau & O’Hara, 2009). However, other studies report either no cross-sectional anxiety-related differences in global cognition within LLD (DeLuca et al., 2005) or no additive effect of anxiety on cognitive dysfunction (Martinussen et al., 2019). These discrepancies may be due to our larger sample with higher statistical power compared to previous work (DeLuca et al., 2005) and the outpatient setting of our data collection compared to the inpatient setting of previous work (Martinussen et al., 2019). Overall, our findings, in the context of other work that demonstrated anxiety as a predictor of future cognitive decline (DeLuca et al., 2005; Fung et al., 2018), suggest that screening for anxiety symptoms in individuals with LLD may be important as high symptom severity may put individuals at risk for cognitive dysfunction and cognitive decline over time. However, it is also possible that cognitive dysfunction leads to increased anxiety.

We also found that higher anxiety was associated with lower OFC volume consistent with previous work in older adults with LLD (Laird et al., 2019) and a history of depression (Potvin et al., 2015), as well as with work in older adults with generalized anxiety disorder outside the context of depression (Andreescu et al., 2017). Lower OFC volume, whether life-long or acquired, may be related to functional disruptions in the OFC that may impact behavioral flexibility (Schoenbaum et al., 2009) and emotion regulation (Dixon et al., 2017) and lead to greater anxiety symptoms. Alternatively, individuals with elevated anxiety may be at risk for atrophy in the OFC, which may contribute to cognitive dysfunction and exacerbate emotion processing difficulties. Future work would benefit from further examining these associations to determine the degree to which smaller OCF volume may be an anxiety-specific biomarker across depressed and non-depressed populations. Unlike previous work (Laird et al., 2019; Potvin et al., 2015), we found no associations between anxiety severity and insula volume. This may be due to differences in analytic approaches. For example, prior work (Laird et al., 2019) did not control for depression severity in their analyses with anxiety; however, in a separate model, the same study found that depression severity was also associated with insula volume in LLD. Thus, lower insula volume may be more specific to depression symptom severity, which we controlled in our models.

Another key finding is that cortical Aβ burden was not associated with anxiety in this LLD sample. To our knowledge, this was the first LLD study to examine Aβ burden in relation to anxiety symptom severity. Our findings suggest that emerging AD, as measured by global cortical Aβ burden, may not be a significant contributor to anxiety symptomatology in this population. These results stand in contrast to previous studies demonstrating associations between anxiety and Aβ burden in non-LLD older adult samples (Banning et al., 2020; Johansson et al., 2022; Krell-Roesch et al., 2018; Lewis et al., 2022), which have been interpreted as anxiety possibly increasing risk for Aβ deposition through elevated cortisol levels and brain inflammation (Banning et al., 2020; Krell-Roesch et al., 2018), or that higher Aβ burden may predispose older adults to developing anxiety (Johansson et al., 2022). It is possible that major depression modulates how the brain responds to anxiety and how anxiety influences the development of cortical Aβ pathology. However, given prior work suggesting links between increased anxiety and high subcortical amyloid burden (Hanseeuw et al., 2020) which was not examined here, it will be important to investigate these associations in future LLD work. We also found no associations between anxiety severity and WMH in our sample consistent with multiple previous non-LLD studies that also failed to demonstrate a link between WMH and anxiety (Andreescu et al., 2017; Clancy et al., 2021; Johansson et al., 2022). These findings suggest that despite consistent links between depression and white matter pathology (Lavretsky et al., 2008; Rensma et al., 2018; Teodorczuk et al., 2007), and limited work showing links between anxious symptomatology and WMH in older adults Kim et al., (2011; Vogels et al., 2007), anxiety may not impart additional WMH risk burden in an already clinically depressed older adult population. Altogether, these results suggest that in LLD the pathophysiology of elevated anxiety symptoms may be primarily related to lower OFC volume and not to emerging AD or CVD pathology. However, these results do not prove a cause-effect relationship. Associations may stem from life-long lower OFC volume that predisposes individuals to anxiety. Longitudinal studies are needed to clarify whether lower OFC volume represents atrophy and if volume loss results from anxiety symptoms or whether OFC atrophy increases the risk for anxiety in LLD.

In addition to global cognition and indicators of neurodegeneration and neuropathology, we also examined whether anxiety severity is associated with functional ability and found no associations in the present sample. Although this finding was unexpected, our findings do align with some previous studies examining anxiety in LLD that also found no association between anxiety and functional ability (Jeste et al., 2006; Lenze et al., 2000), but not with others that did demonstrate a link (Morin et al., 2020; Schoevers et al., 2003). It is also possible that our ability to identify an association was limited because our LLD sample exhibited mostly mild levels of functional impairment, as individuals with dementia, where functional impairment is more prevalent, were excluded. As such, it is essential to investigate the temporal relationship between anxiety in LLD and declines in functional ability over time, given longitudinal evidence that anxiety and depression symptoms are associated with risk for functional decline (Dong et al., 2020; Krell-Roesch et al., 2020; Norton et al., 2012), especially when symptoms are comorbid (Kang et al., 2017).

Finally, a global model examining the present study’s key findings revealed that among the two factors identified as significantly associated with anxiety severity in initial analyses (i.e., global cognitive functioning and OFC volume) only global cognitive functioning remained associated with anxiety severity, independent of OFC volume and concurrent depression severity. To our knowledge, this is the first LLD study to examine anxiety in the context of its relative associations with neurodegenerative features and our findings suggest that global cognitive dysfunction is most salient to anxiety in LLD among the factors examined here. However, this does not diminish our initial finding that anxiety is associated with OFC volume. A relatively restricted range of OFC volumes and limited sample size may have limited our ability to detect a small significant OFC effect once global cognitive functioning was in the model. Global cognitive functioning is associated with OFC integrity and also represents broader aspects of brain functioning that rely on neural integrity beyond regions assessed in the present study. Thus, OFC volume may not have remained significantly associated with anxiety in the global model because variance associated with OFC volume was subsumed by global cognitive functioning. Nevertheless, our finding that OFC volume is associated with anxiety in initial analyses remains of interest as it suggests that OFC volume is one neural factor that is relevant to anxiety severity in LLD and therefore warrants further investigation in future studies. It is also important for future work to examine additional measures of neural integrity and their association with anxiety severity in LLD, using functional neuroimaging tools for example. Effect sizes of the statistically significant associations identified in this study were small to medium in magnitude which sets the stage for further investigations evaluating the theoretical and clinical relevance of these effects.

The association between anxiety and cognitive dysfunction may be bidirectional, such that increased anxiety leads to poorer cognition and that poorer cognition could lead to increased anxiety, which is supported by prior work (Petkus et al., 2017). Possible clinical implications of this finding are that LLD interventions that address symptoms of anxiety in addition to depression may lead to improved cognitive outcomes and that addressing cognitive dysfunction in interventions could decrease anxiety. Multi-pronged prevention and intervention approaches that address both depression and anxiety symptoms and additionally bolster cognitive functioning may be of value.

It is important to point out that the present cross-sectional study did not examine changes over time, and it is possible that both anxiety and cognitive dysfunction are caused by neurodegenerative changes that were not assessed. The cross-sectional nature of this work also precludes conclusions about risk for future atrophy, neuropathology, and cognitive or functional impairment. Longitudinal studies are needed, especially in light of evidence that anxiety symptoms in mild cognitive impairment have been linked to increased risk for conversion to AD (Mah et al., 2015). Examining the temporal relations among anxiety, cognition, and neurodegeneration in longitudinal LLD work can further inform treatment of anxiety symptoms and cognitive dysfunction in LLD. Another limitation of the present study is that limited power in our sample may have contributed to our null findings concerning associations between anxiety severity and cortico-limbic volumes other than the OFC, Aβ deposition, white matter lesion burden, and functional ability. Future work with larger samples is needed to further evaluate these associations.

Conclusion

In summary, we found that lower OFC volume may play a role in anxiety symptoms in LLD more so than neuropathological indicators of emerging AD or CVD. Our results further indicated that cognitive dysfunction is a key factor associated with anxiety severity in LLD. Implications for clinical practice include that the potentially bidirectional relationship of anxiety symptoms with cognitive dysfunction may be responsive to cognitive or psychiatric interventions.