Cognitive function in individuals with and without painful and painless diabetic polyneuropathy—A cross‐sectional study in type 1 diabetes

Abstract Introduction Previous studies suggest that cognitive impairment is more prevalent in individuals with painful and painless diabetic peripheral neuropathy (DPN). However, the current evidence is not well described. This study investigated cognitive function in adults with type 1 diabetes mellitus (T1DM) and the association to painful/painless DPN and clinical parameters. Methods This cross‐sectional, observational, case–control study included 58 participants with T1DM, sub‐grouped into 20 participants with T1DM and painful DPN, 19 participants with T1DM and painless DPN, 19 participants with T1DM without DPN, and 20 healthy controls were included. The groups were matched for sex and age. The participants performed Addenbrooke's examination III (ACE‐III), which assesses attention, memory, verbal fluency, language and visuospatial skills. Working memory was evaluated using an N‐back task. Cognitive scores were compared between the groups and correlated to age, diabetes duration, HbA1c and nerve conduction measurements. Results Compared to healthy controls, T1DM participants showed lower total ACE‐III (p = .028), memory (p = .013) and language scores (p = .028), together with longer reaction times in the N‐back task (p = .041). Subgroup analyses demonstrated lower memory scores in those with painless DPN compared with healthy controls (p = .013). No differences were observed between the three T1DM subgroups. Cognitive scores and clinical parameters were not associated. Conclusions This study supports the notion of cognitive alterations in T1DM and indicates that cognitive function is altered in T1DM regardless of underlying neuropathic complications. The memory domain appears altered in T1DM, particularly in those with painless DPN. Further studies are needed to verify the findings.


| INTRODUC TI ON
Cognitive function has been suggested to be mildly to moderately affected in individuals with type 1 diabetes mellitus (T1DM). [1][2][3] Clinically significant cognitive impairment in 28% of a cohort with middle-aged T1DM participants with childhood-onset diabetes was found, while a prevalence of 5% was observed in healthy controls. 4 The underlying mechanisms are unknown, but it has been suggested that cognitive impairments in T1DM may be attributed to several diabetes-related factors including higher HbA1c, retinopathy and diabetic peripheral neuropathy (DPN). 4,5 DPN is the most common complication of T1DM, 6 affecting up to 50% of all diabetic individuals. 7,8 It presents with loss of sensitivity in lower extremities 6 and can have severe consequences due to an elevated risk of developing foot ulcers, which may lead to amputations and premature death. 9,10 In one study, individuals with T1DM and mild or clinically relevant cognitive impairment had a higher DPN prevalence than T1DM participants who presented with normal cognitive function. 4 Another study demonstrated a correlation between decreased nerve conduction velocity and decreased cognitive function in individuals with T1DM, indicating a possible link between changes in the peripheral and central nervous systems. 5 Also, specific cognitive domains have been reported affected in T1DM with DPN including psychomotor speed and attention. 11 Another factor associated with cognitive impairment is chronic pain conditions, where problems with concentration, memory, processing and attention are prevalent. 12,13 A common cause of chronic pain is peripheral neuropathic pain, a condition that has shown changes in the neural pathways involved in cognition. 9,14 Peripheral neuropathic pain is experienced by 15-25% of individuals with diabetes and DPN in their lower extremities and is known as painful DPN. 9 Considering that diabetes, painless DPN and painful DPN appear to be related to cognitive impairment a combination of these may increase the risk of cognitive impairment. Limited attention has been placed on understanding the underlying mechanisms of T1DM-related cognitive impairment and the impact of factors such as painful and painless DPN on cognitive function. This is necessary to prevent further cognitive decline and offer appropriate supportive measures. The present study hypothesized that individuals with T1DM in general would present lower cognitive scores compared to healthy controls and that painful and painless DPN contribute to Furthermore, associations between impaired cognitive function in T1DM and several disease characteristics have been observed in other studies. 3,15,16 However, their impact on cognitive function has still not been fully elucidated. The final aim was therefore to explore associations between cognitive scores and clinical parameters, including age, diabetes duration and HbA1c. Parameters reflecting DPN such as nerve conduction measurements were also included. ACE-III, cognitive function, diabetic peripheral neuropathy, memory, N-back task, neuropathic pain, type 1 diabetes Exclusion criteria for participants with T1DM and healthy controls included previous or current alcohol and/or drug abuse, presence of chronic viral infection, known neural damage or disease in the neural system or critical ischemia of the lower extremities; severe skin disease; pregnancy, active cancer-disease and previous chemotherapy or consumption of experimental medicine. All participants provided informed written consent before trial enrollment.

| Study design and participants
The study was conducted according to the Declaration of Helsinki.

| Subgrouping and clinical parameters
Painful DPN was clinically confirmed by two independent medical doctors and supported by the self-reporting DN4 questionnaire. A score of 4 or above was considered abnormal, and these participants were classified as having painful DPN. 21 Painless DPN was confirmed according to the Toronto consensus. 7 Thus, those with abnormal vibration perception threshold Other obtained clinical parameters included retinopathy status, which was diagnosed in accordance with the local clinical standards of Department of Ophthalmology, Aalborg University Hospital, Denmark, where the minimum diagnosis criteria for retinopathy were retinal microaneurysms. 22 Due to the risk of excessive glucose levels in T1DM participants, which may affect cognitive performance, actual glucose levels were obtained before performing the cognitive task. Furthermore, blood samples were taken from all participants to measure HbA1c.

| Addenbrooke's examination
The participants performed Addenbrooke's Cognitive Examination-III (ACE-III), a validated cognitive examination that assesses the following five cognitive domains: attention (18 points), memory (26 points), verbal fluency (14 points), language (26 points) and visuospatial abilities (16 points). 23 Attention was assessed by inquiring about the date, recalling three words and performing serial subtraction. Memory was tested by recalling the three previously repeated words, a fictional name and address, and historical facts. Fluency was evaluated by participant generating words starting with a specific letter and naming animals. Language was tested through completing physical commands using a pencil and paper, writing sentences, repeating polysyllabic words and proverbs, identifying objects in line drawings, and reading irregular words. Visuospatial abilities were assessed by copying diagrams, drawing a clock face, counting dots and recognizing fragmented letters. Further details on ACE-III test can be found elsewhere. 23,24 The total possible score is 100, and higher scores indicate better cognitive function. Several studies have proposed two cut-off scores of 82 and 88 points to screen for cognitive impairment. 25 The lower threshold was used for this study to ensure the presence of cognitive impairment.

| N-back task
The participants performed a visual N-back task, which tests a person's ability to temporary storage, update, manipulation of remembered information and respond to a stimulus. In other words, the task tests the cognitive domains, working memory and psychomotor speed. 26 During the task, a sequence of letters was presented one by one. For each letter, the participant had to decide if the letter was presented N letters previously. The higher number of N, the more difficult the task is. Before performing the task, the participants d′ reflects the sensitivity of the participants to discriminate target letters from distraction letters. Thus, a high d′ indicates that the target is easily detected. 27 The RT reflect the psychomotor performance, thus how quickly the participants react to the visual stimuli in the N-back task with a motor activity (pressing the button).
However, the raw RT value is associated with the speed-accuracy trade-off problem, meaning that decisions are made slowly with high accuracy or fast with a high error rate. 26,28,29 To overcome this problem, the RT was adjusted for accuracy generating weighted RTs (RTW) using the following formula: RTW = RT + (RT (1 − Accuracy)), where accuracy was hits/targets. 26

| Statistical analyses
Demographical, clinical, ACE-III and N-back data were tested for normal distribution using the Shapiro-Wilk test and Q-Q plots.
Depending on the distribution, independent t-tests or Mann-Whitney U tests were performed to compare the T1DM group and healthy controls. Sex, retinopathy status and medications were compared using χ 2 -or fisher's exact tests. For the comparisons across subgroups, one-way ANOVAs and Kruskal-Wallis tests were used.
Post-hoc tests with Bonferroni corrections for multiple comparisons were used. Spearman's correlations were performed to examine associations between cognitive scores and clinical parameters.
Cognitive scores which showed significant differences between the tested groups were chosen for the correlation analysis. Data are presented as mean ± standard deviation or median (interquartile range).

| RE SULTS
All 58 participants included in the current cognitive study completed the cognitive tasks and were included in the reported results. Due to impaired vision, one participant did not complete parts of the cognitive tasks which required visual attention therefore this participant was only included in parts of the analysis. Tables 1 and 2 summarize the demographic and clinical data together with peripheral measurements, neuropathic pain assessments and cognitive scores for the overall T1DM group versus healthy controls and the subgroups versus healthy controls respectively.

| Demographics
Overall, there were no differences when comparing sex, age and body mass index (BMI) between the T1DM group and the control group (all p > .056). As expected, healthy controls had a lower HbA1c value compared to T1DM participants (p < .001). Also, the T1DM group demonstrated a higher prevalence of retinopathy and poorer nerve conduction measurements as well as higher pain scores (all p ≤ .001). Higher use of neuropathic pain medications was observed in the T1DM group (p < .009). See Table 1.
There were no differences when comparing sex, age and BMI between the four subgroups (all p > .299). The diabetes subgroups did not differ in regard to the age of diabetes onset, diabetes duration and glucose level (all p > .053). Healthy controls had a lower HbA1c concentration compared to the three T1DM subgroups (all p < .001).
Additionally, the participants with T1DM and painless DPN had higher HbA1c concentrations than those with T1DM without DPN (p = .012). All three T1DM subgroups had higher prevalences of retinopathy compared to the healthy controls (p < .001). The participants with T1DM and painful DPN and those with T1DM and painless DPN had significantly poorer peripheral nerve conduction measurements compared to the participants with T1DM without DPN and the healthy controls (all p < .019). Moreover, the pain score and use of neuropathic pain medications were higher in the group with painful DPN compared to the other three groups (all p < .001). See Table 2.

| Cognitive alterations in the overall type 1 diabetes group
Overall, participants with T1DM demonstrated a lower total ACE-III score compared to healthy controls (p = .028). Moreover, the memory and language scores were lower in the T1DM group compared to healthy controls (all p < .028) (Figure 1). No differences were observed for the other cognitive domains, including attention, verbal fluency and visual and spatial skills (all p > .237). Twenty-one percent of the participants with T1DM and 5% of the healthy controls had a total score below 82. See Table 1.
In the N-back task, the participants with T1DM demonstrated longer RTs and RTWs during the 0-back block compared to healthy controls (p < .041). However, no differences in RTWs were observed during 1-back and 2-back blocks (all p > .128, see Table 1). The RTWs increased as the memory load of the N-back task increased for both the T1DM group and the control group (p < .001). The index d' did not differ between the two groups (all p > .063).

| Cognitive alterations in subgroups of type 1 diabetes
When comparing ACE-III in the subgroups, an overall difference was observed for the memory domain (p = .022), as shown in Table 2.
The post-hoc analysis revealed a lower memory score in the participants with T1DM and painless DPN compared with healthy controls (p = .026), while no differences were observed between the other subgroups (all p > .096), see Figure 2. No differences were found for the total ACE-III score or in the other domains (all p > .060).
The subgroups did not differ when comparing any parameters of the N-back task including d', RTs, or RTW (all p > .081), see Table 2.

| DISCUSS ION
This study characterized cognitive function in individuals with T1DM compared to healthy controls and in well-phenotyped T1DM subgroups with painful DPN, painless DPN and without DPN. Overall, participants with T1DM had significantly lower total, memory and language ACE-III scores compared to healthy controls. Furthermore, the T1DM group had longer raw and accuracy-weighted reaction times during the 0-back task compared to healthy controls. No correlations were found between cognitive scores and clinical parameters in participants with T1DM. However, older age was associated with lower language scores. When comparing the subgroups, the participants with T1DM and painless DPN revealed a lower memory score than healthy controls.
The lower total ACE-III score observed amongst the individuals with T1DM compared to the healthy controls coincides with our hypothesis and previous findings. 2,4 When applying a threshold of 82 points, 21% of the participants with T1DM met the criteria for cognitive impairment, while the prevalence was 5% for the healthy controls. Similar results were found in a study by Nunley and coworkers, where mild cognitive impairment was identified in 28% of individuals with T1DM and 5% of the healthy controls. 4 Their T1DM group also included individuals with different types of diabetic complications.
When comparing subgroups, we found no significant differences in the total ACE-III score, suggesting that T1DM could be a general risk factor for cognitive impairment regardless of the presence of painful and painless DPN.
When examining the different cognitive domains, the individuals with T1DM had lower memory scores compared to healthy controls, suggesting that memory was affected in the T1DM participants. In the subgroups, the individuals with T1DM and painless DPN demonstrated lower memory scores compared to healthy controls, which could indicate that diabetic neuropathy is not only present in the peripheral nervous system but is also present centrally, and could thus have an influence on impaired memory function. This is in line with previous studies which suggest that peripheral nerve damage may serve as a risk factor for cognitive impairment. 4

| Strengths and limitations
The main strength of the present study was the inclusion of three However, this factor did not differ between groups as they were matched for age. damage, microvascular disease, chronic pain, etc., or a combination of disease characteristics that may be linked to the development of DPN and cognitive impairment respectively. Individuals with painless DPN seem to be more likely to present with impaired memory.

| CON CLUS ION
However, due to our small sample size, we cannot draw definite conclusions and larger longitudinal studies in well-characterized T1DM groups with larger sample sizes are needed to confirm our results.
This explorative study provides relevant insights about the cognitive changes in T1DM, regardless of underlying neuropathic complications. This study also indicates neuropathy as a potential contributing factor to the cognitive changes seen in individuals with T1DM. The Conceptualization (equal); funding acquisition (equal); methodology (supporting); resources (lead); supervision (equal); writing -review and editing (equal). Niels Ejskjaer: Conceptualization (equal); funding acquisition (equal); methodology (supporting); resources (lead); supervision (equal); writing -review and editing (equal).

ACK N O WLE D G E M ENTS
None.

FU N D I N G I N FO R M ATI O N
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

CO N FLI C T O F I NTE R E S T S TATE M E NT
None of the authors have potential conflicts of interest to be disclosed. All authors have approved the final version of the article.

DATA AVA I L A B I L I T Y S TAT E M E N T
Some or all data sets generated during and/or analysed during the current study are not publicly available but are available from the corresponding author upon reasonable request.