Associations of Peripheral Neuropathy Defined by Monofilament Insensitivity with Mild Cognitive Impairment and Dementia in Older Adults

Abstract

Introduction:

The aim of this study is to assess the association of peripheral neuropathy(PN) as defined by monofilament insensitivity with mild cognitive impairment(MCI) and dementia in older adults with and without diabetes.

Methods:

We conducted a cross-sectional analysis of 3,362 Black and White participants in the Atherosclerosis Risk in Communities Neurocognitive Study(ARIC-NCS) who underwent monofilament testing at visit 6 (2016–2017, age 71–94 years). Participants’ cognitive status was classified by an adjudication committee as cognitively normal, MCI, or dementia after completing a comprehensive battery of neurocognitive assessments. We used logistic regression to evaluate the association of PN with MCI or dementia overall and stratified by diabetes status after adjusting for traditional dementia risk factors. We also compared age-adjusted brain MRI measures among a subset (N=1095) of participants with versus without PN.

Results:

Overall, the prevalence of MCI (21.9% vs. 16.7%) and dementia (7.8% vs. 3.9%) were higher among participants with vs. without PN (both P<0.05). After adjustment, PN was positively associated with MCI or dementia in the overall study population (OR 1.45, 95%CI 1.23, 1.73). Results were similar by diabetes status (diabetes: OR 1.38, 95%CI 1.03–1.87; no diabetes: OR 1.48, 95%CI 1.20–1.83; P-for-interaction=0.46). Age-adjusted total and lobar brain volumes were significantly lower in participants with vs. without PN (both, P<0.05).

Discussion/Conclusions:

PN as defined by monofilament insensitivity was associated with cognitive status independent of vascular risk factors and regardless of diabetes status. Our findings support a connection between PN and cognitive impairment, even in the absence of diabetes.

Introduction

Peripheral neuropathy (PN) is a major complication of diabetes [1]. However, decreased lower extremity sensation, a common symptom of PN, is also highly prevalent among older individuals (>70 years) without diabetes, affecting over 30% of adults in the general population [2].

Although there are several types, traditionally PN is thought to be caused by microvascular damage primarily involving the distal portions of the longest sensory axons, thereby leading to decreased sensation in the distal lower extremities before affecting other nerves [3]. As a result, PN is one of the manifestations of type 2 diabetes since the hyperglycemia from long-standing diabetes can cause alterations in the nerve structure. We have previously demonstrated that PN as defined by monofilament insensitivity is associated with other microvascular complications, including erectile dysfunction [4], hearing and vision impairment (Hicks et al., Under Review), biomarkers of kidney disease, and elevated cardiac troponin-T [5]. Microvascular disease has also been implicated in the pathophysiology of cognitive disorders including the entity known as Vascular Contributions to Cognitive Impairment and Dementia (VCID) and, more recently, Alzheimer’s disease [6]. In addition to its importance in PN, type 2 diabetes is a robust risk factor for cognitive decline and dementia in older adults [7–11], and is associated with smaller brain volumes and an increased burden of brain vascular pathology [12]. While it may be that shared risk factors contribute to both the development of PN and cognitive decline, the association of PN with cognitive function is not well characterized, nor is their association in individuals without diabetes.

The aim of our study is to assess the association of PN as defined by monofilament insensitivity with a diagnosis of mild cognitive impairment (MCI) or dementia, specific measures of cognitive function, and brain magnetic resonance imaging (MRI) findings among older adults with and without diabetes. We hypothesized that adults with PN will have a higher prevalence of MCI or dementia, lower scores on cognitive tests, and evidence of small vessel disease as well as smaller regional volumes on brain MRI.

Methods

The Atherosclerosis Risk in Communities (ARIC) Study is an ongoing community-based prospective cohort study of 15,792 adults enrolled between 1987 and 1989 (study visit 1, ages 45–64 years) from four US communities (Forsyth County, North Carolina; Jackson, Mississippi; suburbs of Minneapolis, Minnesota; and Washington County, Maryland). ARIC Study participants have been followed longitudinally with serial in-person exams and (semi)annual phone interviews. As part of the ARIC Neurocognitive Study, ARIC participants underwent cognitive testing at visits 2 (1990–1992), visit 4 (1996–1999), visit 5 (2011–2013), and visit 6 (2016–2017). The ARIC study was approved by the Institutional Review Boards at each study site and informed consent was obtained from all participants.

We conducted a cross-sectional analysis of black and white participants in the ARIC Neurocognitive Study who underwent standard monofilament testing as well as cognitive testing at visit 6 (2016–2017). Among the 4,003 black and white participants who attended ARIC visit 6, we excluded participants who were missing monofilament testing (N=386), cognitive function data (N=34), or any important covariates of interest (N=221), leaving 3,362 participants included in the present primary analysis.

Brain MRI data were not available at ARIC visit 6. We therefore assessed the association of PN with markers of neurodegeneration on brain MRI performed at ARIC visit 5 (2011–2013) [12]^,[13]. A subset (N = 1095) of ARIC participants without contraindications were offered a brain MRI at visit 5, including participants enrolled in the ARIC Brain MRI Ancillary Study [14, 13], ARIC-Neurocognitive Study participants with low cognitive test scores and/or evidence of longitudinal decline, and an age-stratified random sample of participants without evidence of cognitive impairment [15].

Exposure of Interest

Our primary exposure of interest was PN, as defined by monofilament insensitivity. Peripheral neuropathy data were collected at using Semmes-Weinstein 10 g monofilament testing at ARIC visit 6. Monofilament testing was applied to the plantar-hallux, plantar-first metatarsal head, plantar-third metatarsal head, and plantar-fifth metatarsal head of each foot. Each site was tested three times by a certified technician and modeled after the NHANES protocol [16]. If two of three responses for a site were incorrect, that site was defined as being insensate. Peripheral neuropathy was defined as having at least one insensate site.

Covariates

Important covariates of interest included sociodemographic characteristics (age, sex, race, education), body mass index [BMI]), lifestyle factors (smoking status, alcohol consumption), comorbidities (diabetes, cardiovascular disease, stroke, cancer, hypertension, hypercholesterolemia, peripheral artery disease [17]), and APOE e4 genotype. Sociodemographic variables were ascertained at ARIC visit 1. All other variables were ascertained at ARIC visit 6. Specific definitions of covariates are summarized in the Supplementary Material.

Outcomes

The primary outcome of interest was the presence of MCI or dementia at visit 6. At ARIC visits 5 and 6, participants were classified by an adjudication committee as cognitively normal or having MCI or dementia based on the National Institute on Aging–Alzheimer’s Association workgroup criteria [18, 19] and Diagnostic and Statistical Manual of Mental Disorders, 5th Edition [20] after completing a comprehensive battery of cognitive, neurologic, and behavioral assessments, as previously described in ARIC [21].

Secondary outcomes of interest included cognitive performance on the domains of executive function, language, and memory; global cognitive performance; and measures of markers of subclinical cerebrovascular disease as well as brain volumetric measurements on brain MRI in the subset of ARIC visit 5 participants who underwent brain MRI scans.

The executive function domain of cognitive performance is composed of the Trail-making test parts A and B, digit symbol substitution, and digit span backwards tests. The language domain is composed of the word fluency, animal fluency, and Boston naming tests. The memory domain is composed of the delayed word recall, logical memory, and incidental learning tests. Use of these cognitive domains to assess cognitive function has been previously described in ARIC [15]. We constructed Z-scores for each of the three cognitive domains by averaging the scores of tests within each domain, subtracting the domain mean, and dividing by the domain standard deviation. We also generated a global Z-score by averaging the above three Z-scores.

Brain MRI scans were performed with four 3T scanners (Maryland: Siemens Verio; North Carolina: Siemens Skyra; Minnesota: Siemens Trio; Mississippi: Siemens Skyra; Siemens Medical Solutions, Malvern, Pennsylvania) using a common set of sequences (Supplementary Material). Brain volumes were measured using the MP-RAGE sequences with Freesurfer image analysis software [22, 23]. White matter hyperintensity volume was assessed on T2 FLAIR sequences and measured quantitatively using a previously-described algorithm [24]. Lacunar infarcts were assessed on T2 FLAIR sequences and defined as subcortical lesions in the caudate, lenticular nucleus, internal capsule, thalamus, brainstem, deep cerebellar white matter, centrum semiovale, or corona radiate with central hypointensity >3 mm and hyperintensity ≤20 mm in maximum dimension [25, 26]. Cortical infarcts were assessed T2 FLAIR sequences and defined as lesions measuring >20 mm in minimum diameter [27]. Lobar and subcortical microhemorrhages were assessed on T2*GRE sequences and defined as lesions measuring ≤5 mm in maximum diameter [25].

Statistical Analysis

We examined the prevalence of each of the cognitive outcomes according to PN and diabetes status at ARIC visit 6. We used logistic regression to assess the association of PN with MCI or dementia (combined outcome). Model 1 adjusted for age, sex, and race. Model 2 further adjusted for education, BMI, smoking status, drinking status, diabetes status, cardiovascular disease, stroke, hypertension, hypercholesterolemia, cancer, and APOE status. We also adjusted for diabetes status in the overall model (Model 2a) and duration of diabetes (< 10 years vs. ≥ 10 years) in participants with diabetes (Model 2b). In a secondary analysis, we used multinomial logistic regression to calculate the relative risk ratio (RRR) for the association of PN with MCI or dementia, considered separately, overall and according to diabetes status.

We conducted a non-concurrent cross-sectional analysis of the association of brain MRI measures at visit 5 (2011–2013) with PN at visit 6 (2016–2017) using t-tests and Chi-squared tests incorporating sampling weights to account for selection into the MRI substudy. Brain volume measures were calculated as a proportion of total intracranial volume and transformed to an age-adjusted Z score (5-year age band).

We performed all analyses using Stata, version 15.1 (StataCorp) with P < 0.05 denoting statistical significance.

Results

There were 3,362 ARIC participants who underwent monofilament and cognitive function testing at ARIC visit 6 (Table 1). Mean age was 79.3±4.7 years, 41.1% were male, and 20.4% were black. Overall, 39.2% of participants (N = 1,318) had evidence of PN, including 35.1% of participants with diabetes and 29.8% of participants without diabetes.

Table 1.

Characteristics of ARIC participants at visit 6 (2016–2017) overall and stratified by the presence or absence of peripheral neuropathy

	Overall (N = 3362)	Peripheral Neuropathy
		No (N = 2044)	Yes (N = 1318)
Mean age, years (SD)	79.3 (4.7)	78.6 (4.3)	80.5 (5.0)
Male	41.1%	31.4%	56.2%
Black	20.4%	20.3%	20.5%
Education
Less than high school	11.1%	9.5%	13.5%
High school or vocational school	41.3%	42.7%	39.2%
Some college and above	47.6%	47.7%	47.3%
Body mass index, kg/m2
<25.0	28.9%	30.8%	25.9%
25.0-<30.0	38.9%	39.7%	37.6%
≥30.0	32.2%	29.5%	36.5%
Smoking status
Never	45.0%	46.1%	43.3%
Former	48.3%	47.4%	49.8%
Current	6.7%	6.5%	6.9%
Drinking status
Never	20.4%	20.5%	20.2%
Former	28.3%	27.1%	30.3%
Current	51.3%	52.4%	49.5%
Diabetes status
No	68.1%	70.2%	64.9%
Yes	31.9%	29.8%	35.1%
Prevalent cardiovascular disease	18.2%	15.1%	23.1%
Prevalent stroke	3.5%	2.9%	4.5%
Prevalent cancer	26.0%	24.5%	28.3%
Hypertension	84.0%	82.2%	86.5%
Hypercholesterolemia	60.5%	59.9%	61.4%
Peripheral artery disease*	5.2%	4.2%	6.7%
APOE e4 present	27.2%	27.3%	27.1%

^*Percentages are out of 2965 participants with ankle brachial index measurements at visit 5

The prevalence of MCI or dementia was significantly higher among participants with vs. without PN (29.7% vs. 20.6%, P<0.001; Table 2). Overall, MCI was present in 21.9% of participants with PN vs. 16.7% of participants without PN (P<0.001). Dementia was present in 7.8% of participants with PN vs. 3.9% of participants without PN (P<0.001). Among adults without diabetes, the prevalence of MCI or dementia was higher among participants with vs. without PN (30.3% vs. 20.0%; P<0.001). Among adults with diabetes, the prevalence of MCI or dementia was also higher among participants with PN (28.5% vs. 22.0%; P=0.014).

Table 2.

Crude (unadjusted) cognitive status measures according to peripheral neuropathy status among ARIC participants at visit 6 (2016–2017)

	Overall			No Diabetes			Diabetes
	No PN (N=2044)	With PN (N=1318)	P-value	No PN (N=1435)	With PN (N=856)	Pvalue	No PN (N=609)	With PN (N=462)	P-value
Cognitive status			<0.001			<0.001			0.035
Normal (N = 2550)	79.4%	70.3%		80.0%	69.7%		78.0%	71.5%
Mild impairment (N = 630)	16.7%	21.9%		16.5%	22.1%		17.4%	21.4%
Dementia (N = 182)	3.9%	7.8%		3.5%	8.2%		4.6%	7.1%
Global z-score, mean (SD)	−0.21 (0.94)	−0.45 (0.98)	<0.001	−0.17 (0.95)	−0.47 (0.98)	<0.001	−0.29 (0.92)	−0.42 (0.98)	0.027
Executive functioning, z-score, mean (SD)	−0.23 (1.25)	−0.64 (1.36)	<0.001	−0.18 (1.23)	−0.63 (1.36)	<0.001	−0.36 (1.29)	−0.66 (1.36)	<0.001
Language, z-score mean (SD)	−0.21 (1.11)	−0.36 (1.19)	<0.001	−0.18 (1.12)	−0.36 (1.20)	<0.001	−0.30 (1.08)	−0.36 (1.18)	0.41
Memory, z-score mean (SD)	−0.18 (1.23)	−0.37 (1.22)	<0.001	−0.17 (1.25)	−0.44 (1.23)	<0.001	−0.20 (1.19)	−0.24 (1.19)	0.61

^aIncludes trail making test parts A and B, digital symbol substitution, and digit span backwards tests

^bIncludes word fluency, animal fluency, and Boston naming tests

^cIncludes delayed word recall, logical memory, and incidental learning tests

Global and executive functioning scores were significantly lower for participants with versus without PN overall and among adults with and without diabetes (Table 2). Language and memory scores were lower for participants with PN overall (both, P<0.001), but this was primarily driven by adults without diabetes. In contrast, language (P=0.41) and memory scores (P=0.61) did not significantly differ according to the presence or absence of PN among participants with diabetes.

Peripheral neuropathy as defined by monofilament insensitivity was positively associated with MCI or dementia (OR 1.47, 95% CI 1.24–1.73) in the overall study population (Model 1; Table 3). This association persisted after adjusting for additional risk factors (OR 1.46, 95% CI 1.23–1.73; Model 2a). When participants were stratified by diabetes status, the adjusted associations of PN with MCI or dementia were similarly robust for adults with (OR 1.40, 95% CI 1.03–1.89) and without (OR 1.47, 95% CI 1.19–1.82) diabetes, with no evidence for interaction by diabetes status (P=0.46 for interaction in Model 1).

Table 3.

Association (OR, 95% CI) of peripheral neuropathy with cognitive impairment (mild cognitive impairment or dementia) overall and stratified by diabetes status, ARIC visit 6 (2016–2017)

	N cases / N	Cognitive Impairment OR (95% CI)
		Model 1	Model 2
Overall
No Peripheral Neuropathy	421/2044	1 (Ref)	1 (Ref)
Peripheral Neuropathy	391/1318	1.47 (1.24–1.73)	1.46 (1.23–1.73)*
No Diabetes
No Peripheral Neuropathy	287/1435	1 (Ref)	1 (Ref)
Peripheral Neuropathy	259/856	1.49 (1.21–1.83)	1.47 (1.19–1.82)
Diabetes
No Peripheral Neuropathy	134/609	1 (Ref)	1 (Ref)
Peripheral Neuropathy	132/462	1.40 (1.05–1.87)	1.40 (1.03–1.89)**

Model 1: Adjusted for age, sex and race.

Model 2: Further adjusted for education, body mass index, smoking status, drinking status, diabetes status, cardiovascular disease, stroke, hypertension, hypercholesterolemia, cancer and APOE.

^*Model 2a = additionally adjusted for diabetes status

^**Model 2b = additionally adjusted for duration of diabetes (< 10 yrs vs. ≥ 10 yrs)

Peripheral neuropathy was also positively associated with the separate outcomes of MCI (RRR 1.37, 95% CI 1.14–1.65) and dementia (RRR 1.85, 95% CI 1.35–2.55) in the overall study population (Model 1; Table 4). These associations persisted after adjusting for additional risk factors (Model 2a). When participants were stratified by diabetes status, the adjusted associations of PN with MCI (P=0.64 for interaction) and dementia (P=0.29 for interaction) were similar for adults with and without diabetes, although the association of PN with dementia among adults with diabetes was no longer significant (RRR 1.38, 95% CI 0.78, 2.44), likely due to low power in this subgroup.

Table 4.

Adjusted relative risk ratio ratios (RRR, 95%CIs) for mild cognitive impairment and dementia according to the presence/absence of peripheral neuropathy (PN) overall and stratified by diabetes status, ARIC visit 6 (2016–2017)

	Mild Cognitive Impairment vs. Normal Cognition			Dementia vs. Normal Cognition
	N cases / N	RRR (95% CI)		N cases / N	RRR (95% CI)
		Model 1	Model 2		Model 1	Model 2
Overall
No PN	342/2044	1 (Ref)	1 (Ref)	79/2044	1 (Ref)	1 (Ref)
With PN	288/1318	1.37 (1.14–1.65)	1.38 (1.14–1.67)*	103/1318	1.85 (1.35–2.55)	1.83 (1.32–2.55)*
No Diabetes
No PN	236/1435	1 (Ref)	1 (Ref)	51/1435	1 (Ref)	1 (Ref)
With PN	189/856	1.38 (1.10–1.73)	1.36 (1.08–1.72)	70/856	1.95 (1.32–2.90)	2.07 (1.37–3.12)
Diabetes
No PN	106/609	1 (Ref)	1 (Ref)	28/609	1 (Ref)	1 (Ref)
With PN	99/462	1.34 (0.97–1.85)	1.41 (1.02–1.97)**	33/462	1.60 (0.93–2.74)	1.38 (0.78–2.44)**

Model 1: Adjusted for age, sex and race.

Model 2: Further adjusted for education, body mass index, smoking status, drinking status, diabetes status, cardiovascular disease, stroke, hypertension, hypercholesterolemia, cancer and APOE.

^*Model 2a = additionally adjusted for diabetes status

^**Model 2b = additionally adjusted for duration of diabetes (< 10 yrs vs. ≥ 10 yrs)

In the subset of participants with brain MRI, age-adjusted total brain volume was lower for participants with vs. without PN (P=0.004; Table 5). Participants with PN also had lower age- and total brain-volume adjusted lobar volumes for the frontal, temporal, occipital, and parietal lobes (P≤0.009). There were no significant differences in age- and total-brain volume adjusted white matter hyperintensity volumes or the prevalence of brain microhemorrhages or infarcts for participants with PN compared to those without (P>0.05; Table 5).

Table 5.

Brain MRI measures (at ARIC visit 5)^* according to peripheral neuropathy status among ARIC participants at visit 6 (2016–2017)

	Peripheral Neuropathy		P-value
	No (Unweighted N = 657)	Yes (Unweighted N = 438)
White matter hyperintensity volume, age-adjusted z-score (SE)**	−0.05 (0.09)	0.08 (0.13)	0.38
Lobar microhemorrhages	5.4%	7.2%	0.32
Subcortical microhemorrhages	15.3%	19.8%	0.11
Cortical infarcts (large+small+subcortical)	18.8%	20.6%	0.55
Lacunar infarcts, <20mm	13.0%	13.6%	0.82
Total brain volume, age-adjusted z-score (SE)***	0.10 (0.06)	−0.11 (0.04)	0.004
Lobar volume (frontal+temporal+parietal+occipital), age-adjusted z-score (SE)***	0.15 (0.04)	−0.18 (0.06)	< 0.001
Frontal volume, age-adjusted z-score (SE)	0.10 (0.05)	−0.18 (0.05)	< 0.001
Temporal volume, age-adjusted z-score (SE)	0.12 (0.04)	−0.17 (0.06)	< 0.001
Parietal volume, age-adjusted z-score (SE)	0.16 (0.05)	−0.04 (0.06)	0.009
Occipital volume, age-adjusted z-score (SE)	0.16 (0.05)	−0.14 (0.05)	< 0.001

^*Weighted mean or percentage

^**Age-adjusted (5-year age band) z-score of white matter hyperintensities volume as a percentage of total “at risk” white matter

^***Age-adjusted (5-year age band) z-score of total brain volume/lobar volume as a percentage of estimated total intracranial volume

Discussion

We found significant and robust cross-sectional associations of PN as defined by monofilament insensitivity with the combined outcome of MCI or dementia among older adults in the ARIC study. The associations were similar for adults with and without diabetes, and predominantly manifested as reduced executive function. On brain MRI, participants with PN had decreased total brain volume and lobar volumes. Together, our data support a possible shared (micro)vascular pathophysiology of PN and cognitive impairment that is independent from diabetes status.

An association of PN with cognition has been described in a few studies limited to adults with diabetes [28, 29]. A small cross-sectional study using functional MRI recently showed that cerebral activity in response to an acute thermal stimulus to the foot is stronger among adults with type 2 diabetes and PN compared to adults with type 2 diabetes and no PN [30]. This activity involved the brain areas that participate in somatosensory pathways (i.e. right insula, left caudate nucleus, frontal gyrus, and cingulate cortex), but also in the cognition-related cerebral areas (i.e. right temporal lobe, left hippocampus, and left fusiform gyrus). A recent small cross-sectional study of 70 adults with type 1 diabetes showed that PN was independently associated with cognitive dysfunction based on cross-sectional assessment with the Montreal Cognitive Assessment [31]. In contrast, a study of 94 adults with type 2 diabetes demonstrated no significant correlation between PN and cognitive function as measured by the Mini-Mental State Examination [32]. These conflicting results may reflect differences in the definition of PN, the cognitive assessments, diabetes duration/severity, and/or a lack of statistical power in these smaller studies. Future studies assessing the association of PN diagnosed with more specific evaluation with cognitive impairment will be important moving forward.

Peripheral neuropathy resulting in decreased lower extremity sensation is a known complication of diabetes, and its prevalence increases with long-standing hyperglycemia [33]. Similarly, diabetes status, poor glycemic control, and diabetes duration have been closely linked to worse cognitive outcomes in older adults [34]. However, the etiology and consequences of PN among adults without diabetes are not well described. Based on data from NHANES, PN (as defined by monofilament testing) affects approximately 11% of the US population over age 40 years without diabetes [17], which is equivalent to 35 million adults based on the 2020 US Census. The prevalence is much higher in older adults [35], consistent with the high (39.2%) prevalence we report in this cohort of patients with a mean age of 79.3 years. Whether nondiabetic PN is a manifestation of microvascular disease or another etiology is unclear. Our finding that PN as defined by monofilament insensitivity is associated with MCI or dementia among adults both with and without diabetes is novel and suggests a shared pathophysiology of disease that is distinct from hyperglycemia.

We have previously shown robust associations of PN as defined by monofilament insensitivity with other manifestations of microvascular disease among adults with and without diabetes [5, 4] It is plausible that microvascular disease may play a common role in the link between PN and cognitive impairment. Based on brain MRI, we found a significant association of PN with smaller brain volumes. Brain atrophy has been association with cerebral small vessel disease in the past [36]. In addition, covert cerebrovascular disease and its consequences on cognitive performance and brain atrophy have been shown to affect clinically asymptomatic persons as young as 50 years old [37]. Early cerebral small vessel disease and lower brain volumes have been associated with alterations in gait as well [37]. Risk factors for brain atrophy are similar to those for PN and include glycemic control and diabetes duration [12]as well as standard risk factors for metabolic syndrome and atherosclerotic disease [38–40]. It would be interesting to evaluate if biomarkers for microvascular disease are shared by patients with PN and cognitive impairment in future studies.

The associations of PN with white matter hyperintensities, lacunar infarcts, or microhemorrhages were not statistically significant in our cohort, although these manifestations of cerebral small vessel disease were all more common in persons with PN in our study. White matter hyperintensities are associated with cerebral microvascular disease, cardiovascular risk factors, and cognitive decline in older adults. [41, 42] Silent brain infarcts have been associated with hypertension, but less so with other cardiovascular risk factors [43]. There is some evidence that cardiovascular risk factors may be associated specifically with lacunar infarcts between 820mm in diameter, representing microatheroma [44]. Our analysis of brain MRI measures was restricted to a subsample of ARIC participants, was non-concurrent (brain MRIs were conducted ∼4 years before the PN assessment), and was limited to the age-adjusted association of PN with any infarct <20mm, which may have been too non-specific to detect meaningful differences between groups.

Subjective cognitive impairment is a common concern among older adults but can represent heterogeneous complaints ranging from sequelae of the normal aging process to clinical depression, anxiety, or memory loss [45]. Screening for subjective cognitive impairment has been proposed as a means for primary care physicians to identify adults at risk for objective cognitive dysfunction who may benefit from neuroimaging or other assessments [46]. Although there was a higher prevalence of MCI or dementia in adults with PN in our study, it is unclear if screening for PN or if cognitive screening in individuals with PN would improve outcomes. However, it is notable that PN, lower cognitive performance, and brain atrophy are associated with gait disturbances [37, 47, 48]. This puts individuals in whom PN, cognitive impairment, and/or brain atrophy co-occur at especially high risk of gait problems, potentially resulting in falls and other morbidity [49]. Understanding these relationships may heighten our awareness of fall risk in older adults, which may not be limited to adults with diabetic peripheral neuropathy.

The limitations of our study include the cross-sectional design of our study, which did not allow us to evaluate the temporal association of PN and cognitive impairment. We used monofilament testing to assess PN because it has been used in population-based studies in the past, is not painful to the patient, and is a valid screening test for severe PN [50]. However, the gold standard diagnosis of PN is with nerve conduction studies [51]. It is possible that patients with cognitive impairment may not have been able to participate in the monofilament exam as accurately as patients with normal cognitive function. It is also possible that patients with advanced dementia may not have attending ARIC visit 6, leading to selection bias. However, this selection bias should have skewed our results to being more conservative. Finally, our analysis of PN and brain MRI findings was based on non-concurrent data collected approximately 3–5 years apart. Strengths of our study included a large cohort size, robust cognitive assessment of all participants with cognitive function determined based on a formal adjudication committee, and standardized monofilament testing for PN using a protocol designed for use in NHANES.

In conclusion, PN as defined by monofilament insensitivity and cognitive status were closely associated in older adults independent of other risk factors. Results were similar in persons with and without diabetes. Our findings support a connection between PN and cognitive impairment, even in the absence of diabetes.