Associations between brain structure and dual decline in gait and cognition

Dual decline in gait and cognition is associated with an increased risk of dementia, with combined gait and memory decline exhibiting the strongest association. To better understand the underlying pathology, we investigated the associations of baseline brain structure with dual decliners using three serial gait speed and cognitive assessments in memory, processing speed-attention, and verbal fluency. Participants (n = 267) were categorized based on annual decline in gait speed and cognitive measures. Lower gray and white matter volume and higher white matter hyperintensity volume increased the risk of being a dual decliner in gait and both the memory and processing speed-attention groups (all p < 0.05). Lower hippocampal volume (p = 0.047) was only associated with dual decline in gait and memory group. No brain structures were correlated with dual decline in gait and verbal fluency. These results suggest that neurodegenerative pathology and white matter hyper-intensities are involved in dual decline in gait and both memory and processing speed-attention. Smaller hippocampal volume may only contribute to dual decline in gait and memory.


Introduction
The prevalence of dementia is projected to increase significantly in the coming decades due to population growth and ageing (Gale et al., 2018;Nichols et al., 2022).It is estimated that the number of people with dementia will reach approximately 152 million globally by the year 2050 (Nichols et al., 2022).Identifying early biomarkers, establishing preventive measures, and developing new neuroprotective interventions hold the potential to alleviate future disease burden (de la Fuente--Fernández, 2006;Grande et al., 2020a;Livingston et al., 2017;Maczurek et al., 2008;Nichols et al., 2022).Recent studies have provided evidence that dual decline in gait and cognitive function is linked to an elevated risk of developing dementia (Collyer et al., 2022;Grande et al., 2020b;Montero-Odasso et al., 2020;Tian et al., 2023Tian et al., , 2020)), emphasizing the potential of incorporating gait speed into the diagnostic evaluation to enhance early detection (Grande et al., 2020b).Associations between gait speed and various cognitive domains (Martin et al., 2013), along with a wide array of dementia risk factors such as diabetes (Ahsan et al., 2023;Jayedi et al., 2024), hypertension (Ahsan et al., 2023;Shang et al., 2021;Walker et al., 2019), and physical activity (Iso-Markku et al., 2022;Willey et al., 2017), underscore its significance as a valuable measure in dementia risk assessments.However, current understanding of the factors underlying dual decline in gait and cognition is limited.
Interestingly, the combination of dual decline in gait and memory (delayed recall) appears to show the strongest increase in risk of dementia, compared to the combination of gait and other cognitive domains (Collyer et al., 2022).This may be because gait exhibits a stronger correlation with non-amnestic domains such as executive function and processing speed (Martin et al., 2013), and when combined with memory this combination captures a broader assessment of cognitive function, and potentially a wider range of underlying pathology.Previous studies have reported that cerebral small vessel disease (cSVD) is associated with slower gait speed (Pinter et al., 2017) and non-amnestic cognitive domains (Debette and Markus, 2010), while hippocampal atrophy is strongly associated with memory dysfunction (Den Heijer et al., 2010;Hashimoto et al., 2017).Hippocampal atrophy is also considered an early marker for Alzheimer's disease (Frankó et al., 2013;Rao et al., 2022;Van Der Flier and Scheltens, 2009).However, there is limited understanding of the neuroimaging substrates underlying the association between dual decline in gait and cognition.
Two studies have been conducted studying whether brain structures are associated with dual decline in gait and cognition (Grande et al., 2023;Tian et al., 2021), and only one regarding memory (Tian et al., 2021).In a longitudinal study (n = 391) there were no significant differences in pre-selected regional brain volumes at baseline between non decliners and dual decliners in gait and memory (California Verbal Learning Test -immediate recall) (Tian et al., 2021).However, this study did not investigate other brain pathology such as cSVD.In a recent Swedish study (n = 385) baseline total brain and hippocampal volume, and a higher burden of white matter hyperintensities (WMHs) were associated with dual decline in gait and cognition measured with the Mini-Mental State Examination, which is a global cognitive screening test (Grande et al., 2023).A broad range of dual decline in gait and specific individual cognitive domains have not yet been explored, and other measures of cSVD such as subcortical infarcts, enlarged perivascular spaces and microbleeds have not been studied in this context.
Given that dual decline in gait and memory is strongest in prediction of dementia risk (Collyer et al., 2022), and because most people with dementia have a mixture of brain pathologies (Schneider et al., 2007), our primary aim was to examine associations of a range of baseline MRI vascular and neurodegenerative markers of dementia with dual decline in gait and memory.We hypothesized that measures of neurodegeneration (smaller hippocampal, gray, and white matter volume) and a greater total burden of cSVD would be associated with increased risk of dual decline in gait and memory.Our secondary aims were 1) to examine associations of these brain measures with dual decline in gait and more fundamental non-amnestic domains (processing speed-attention and verbal fluency) and 2) to examine the association between each of the individual cSVD measures (WMHs, subcortical infarcts, microbleeds and enlarged perivascular spaces) and dual decline groups across all the cognitive domains.We hypothesized that a greater total burden of cSVD, smaller gray matter and white matter volume, but not hippocampal volume, would be associated with dual decline in gait and non-amnestic domains, and that a greater burden of each individual measure of cSVD would be associated with dual decline in gait and all cognitive domains (processing speed-attention, verbal fluency, and memory).

Study population
The study sample was derived from the population-based Tasmanian Study of Cognition and Gait (TASCOG), conducted between 2005 and 2012.People aged > 60 years were randomly selected from the Southern Tasmanian electoral roll (Australia) (Callisaya et al., 2010;Srikanth et al., 2009).Participants (n = 431) underwent gait and cognitive assessments on the same day at baseline and two further time points on average 2.5 years apart (Jayakody et al., 2019).MRI brain scans were acquired within one month of the clinic assessments.For this analysis, we used baseline MRI brain measures and gait and cognitive measures from all time points.Those with a history of dementia (n = 3), missing MRI brain scans (n = 49), or without at least two time points of gait (n = 111) or cognitive (n = 1) measurements were excluded, leaving a final sample of 267 participants.Of the 267 included participants, n = 47 completed two assessments with a follow-up period of 2.56 ± 0.45 years, and n = 220 completed three assessments with a follow-up period of 4.60 ± 0.53 years.Fig. 1 shows the flow of participants in the study.Ethical approval was obtained from the Southern Tasmanian Health and Medical Human Research Ethics Committee and informed written consent was obtained from all participants.
Brain volumes (total intracranial volume, gray matter volume, white matter volume, total hippocampal volume) were calculated by the automated pipeline of FreeSurfer 5.3 software (Fischl et al., 2002).WMH volumes were calculated using manual correction of an automated pipeline with adaptive boosting classification applied to FLAIR, T1 and T2-weighted scans as previously described (Beare et al., 2009;Callisaya et al., 2014b).The resulting WMH masks were used in Free-Surfer to correct misclassification.The presence of subcortical infarcts was determined by consensus between two MRI experts using a standard definition of 3-20 mm diameter with a surrounding hyperintense rim (Callisaya et al., 2014b).Brain microbleeds were identified using T2*-weighted gradient echo images and were defined as punctate (<10 mm) homogeneous foci of low signal intensity on T2*-weighted images.Enlarged perivascular spaces were defined as lesions that were either round, oval, or linear in shape, with a smooth margin, and without any mass effect with a signal intensity equivalent to that of cerebrospinal fluid on T2-weighted images and if visible, hypointense without a hyperintense rim on FLAIR images as previously described (Huijts et al., 2013).Burden of WMHs was divided by quartiles into four groups as per our prior studies (Callisaya et al., 2014b).Additionally, the total burden of cSVD was calculated using an ordinal scale ranging from 0 to 4, which reflects the combined presence of WMHs, enlarged perivascular spaces, microbleeds and subcortical infarcts as previously reported (Huijts et al., 2013).Due to low numbers in the three highest burden groups, these were combined for analysis.

Gait assessment
Gait speed was assessed in the same way at all three time points using a computerized 4.6 m GAITRite walkway (GAITRite system, CIR Systems, PA, USA).The active area of the walkway is 61 cm wide and 366 cm long with 48 × 288 sensors, with an 80 Hz data capturing frequency and a temporal resolution of 11 ms.Each participant completed six walks at their preferred pace starting 2 m before and finishing 2 m after the walkway to allow for acceleration and deceleration.Gait speed (cm/s) was measured averaging all six walks.

Cognitive assessment
Several cognitive tests were selected based on prior studies of dual decline in gait and cognition (Collyer et al., 2022).The same battery of cognitive tests was administered at each phase by a neuropsychologist: (i) Verbal memorythe Hopkins Verbal Learning Test-Revised (HVLT-R) (delayed recall) (Lezak, 1995); (ii) Processing speed-attention -Digit Symbol Coding subset of the Wechsler Adult Intelligence Scale-III (Wechsler, 1997); (iii) Verbal fluency -the Controlled Word Association Test (COWAT, using the letters F, A, and S) reflecting executive function (Lezak, 1995).

Other measures
Demographics and medical history were obtained using a selfreported questionnaire.Medical history included diabetes mellitus, hypertension, stroke, ischemic heart disease, depression, smoking history, and arthritis.Height and weight were measured, and Body mass index (BMI) was calculated.Based on neuropsychological test performance, two clinical neuropsychologists reached a consensus on a diagnosis of cognitive impairment if a participant scored ≤ 1.5 SD below the age-, sex-, and education-appropriate norms in at least one test within each domain across five cognitive domains, as previously described (Jayakody et al., 2022).Subsequently, participants were classified as having cognitive impairment if 2 domains were impaired.

Statistical analysis
The following procedure was used to create the dual decline groups.Firstly, change in gait speed and each cognitive measure per year were estimated using the predictions for each individual from random effects linear regression models with both a fixed and random linear effect for time (per year).That is, a different estimate of change is fitted for each individual.These annual changes were then used to rank participants.Participants were then classified into four groups based on tertiles for gait and each cognitive measure: (1) Non decliners: participants belonging to the middle and the lowest tertiles of annual decline in both gait speed and cognition; (2) Cognitive decliners: participants belonging to the highest tertile of annual decline in only the cognitive measure; (3) Gait decliners: participants belonging to the highest tertile of annual decline only in gait speed; (4) Dual decliners: participants belonging to the highest tertile of annual decline in both gait speed and the cognitive measure.This classification procedure was performed three times using gait speed and each individual cognitive measure (memory, processing speed-attention, verbal fluency) separately.These groups were named and defined as per prior studies of dual decliners (Collyer et al., 2022).
Multinomial logistic regression was used to examine the associations between baseline MRI markers and group membership, with non decliners as the reference group.Primary analysis was performed to investigate the associations between baseline total hippocampal volume, gray matter volume, white matter volume and the cSVD burden score with the gait and memory dual decliner group.Secondary analysis examined 1) the associations between the same brain measures and the other gait and cognition (processing speed-attention, verbal fluency) dual decline groups and 2) the associations between each of the individual cSVD measures (WMH in quartiles, subcortical infarcts, microbleeds and enlarged perivascular spaces) and dual decline groups.Total gray matter, white matter, and hippocampal volume were entered as continuous variables in the models, while subcortical infarcts, microbleeds and enlarged perivascular spaces were entered as binary variables (reference group = 0).The cSVD burden (reference = lowest cSVD burden group (burden score = 0)) and WMH (reference group is having the lowest WMH volume in each group) were categorized into three and four groups respectively, in the models.
Models were adjusted for baseline age, sex (male, female), education level (seven categories depending on the highest educational qualification completed) and total intracranial volume (for brain volume measures).In addition, gray matter and white matter volumes were adjusted for individual features of cSVD where significant, as these features could be associated with cortical atrophy (Appelman et al., 2009;Kloppenborg et al., 2012).We conducted two post-hoc additional analyses.For the primary analyses we examined the effect of including other brain measures in the model for those that were significant.We also adjusted the models for vascular medical history (stroke, hypertension, diabetes mellitus, ischemic heart disease).We do this as an additional and not primary analysis as these variables may be antecedents on the pathway to the brain variables.
Interaction terms between age and sex and the exposure variables were included in the models if statistically significant.The statistical analysis was performed using Stata version 17.0 (StataCorp, College Station, TX, USA).

Sensitivity analysis
Due to the attrition in our analysis dataset from the original random sample we conducted a sensitivity analysis applying propensity weights to our models.To do this we used a logistic model with demographic and medical history variables to model the probability of individuals having complete data.Using predictions of the probability from this model we derived weights as the inverse of this probability, so that individuals similar to groups with lower probability of having complete data were weighted more highly.These weights were applied to each of the primary analysis models.

Results
At baseline, the mean age was 70.9 ± 6.7 years and 146 (54.7 %) were male.Baseline characteristics of included and excluded participants are reported in Supplementary Table 1.Table 1 shows the characteristics of participants of the gait and memory group.Supplementary Table 2 and 3 show the participant characteristics for the other gait and cognitive domain groups.

Primary outcome: Gait speed and memory (HVLT-R) decline
Table 2 shows the unadjusted and adjusted associations of baseline brain measures with group membership for decline in gait speed and memory.The RRR for a group is interpreted as the relative risk ratio of being a member of that group versus being in the reference group (no decline) for a one unit increase in the baseline brain measure.In fully adjusted models lower baseline hippocampal (RRR 0.59 per ml; 95 % CI 0.35, 0.99; p = 0.047), gray matter (RRR 0.83 per 10 ml; 95 % CI 0.70, 0.99; p = 0.037) and white matter volume (RRR 0.82 per 10 ml; 95 % CI 0.72, 0.93; p = 0.002) were associated with subsequent higher risk of dual decline group membership.
Total burden of cSVD showed no association with the dual decline group.However, a linear association was found when WMH was modelled as a four-level categorical variable, with a higher burden of WMHs associated with a higher risk of dual decline (RRR 1.55 per ml; 95 % CI 1.04, 2.29; p = 0.030) and gait only decline group membership (RRR 1.57 per ml; 95 % CI 1.12, 2.19; p = 0.009) (Supplementary Table 4).
For the additional analysis, the RRR strengthened slightly for gray matter when adjusted for white matter volume (RRR 0.77 per 10 ml; 95 % CI 0.64, 0.94; p = 0.009) and also for the white matter model adjusted for gray matter volume (RRR 0.81 per 10 ml; 95 % CI 0.71, 0.92; p = 0.001 (changes are <10 %).The addition of WMH to the model for gray matter (RRR 0.83 per 10 ml; 95 % CI 0.70, 0.99; p = 0.033) and white matter volume (RRR 0.82 per 10 ml; 95 % CI 0.72, 0.93; p = 0.002) did not meaningfully change the strength of the association.When hippocampal volume was adjusted for total gray matter volume there was a 13.5 % change in the RRR for hippocampal volume (RRR 0.67 per ml; 95 % CI 0.38, 1.19; p = 0.174).When adjusting for white matter volume, the association slightly reduced (RRR 0.72 per ml; 95 % CI 0.42, 1.24; p = 0.232).The significant association disappeared when adjusting the model for both gray matter and white matter volume (RRR 1.03 per ml; 95 % CI 0.55, 1.92; p = 0.938).
When adjusting for vascular medical history at baseline, there were no meaningful changes observed in the results: point estimates for the adjusted models for hippocampal volume, and white matter moved from 0.59 to 0.65 (10.2 % change), and 0.82-0.83,respectively, while gray  Note.Model 1 adjusted for age, sex, education level with the addition of total intracranial volume for volume measures.cSVD, cerebral small vessel disease; ml, milliliter; RRR, relative risk ratio; CI, confidence interval Interpreting the Relative Risk Ratio (RRR): the RRR values in the table show the risk of belonging to the decliner group in that column relative to the non decliner group, for a one unit increase in the brain volume measure versus no increase.For the categorical cSVD burden exposure, the comparison is between the burden group relevant to that row, versus the lowest burden group.

Table 3
Associations between baseline brain volumes and total burden of cSVD and group membership for gait speed and processing speed-attention (Digit Symbol Coding test).Note.Model 1 adjusted for age, sex, education level with the addition of total intracranial volume for volume measures.cSVD, cerebral small vessel disease; ml, milliliter; RRR, relative risk ratio; CI, confidence interval Interpreting the Relative Risk Ratio (RRR): the RRR values in the table show the risk of belonging to the decliner group in that column relative to the non decliner group, for a one unit increase in the brain volume measure versus no increase.For the categorical cSVD burden exposure, the comparison is between the burden group relevant to that row, versus the lowest burden group matter remained unchanged at 0.83 in both models (changes are <10 %), and corresponding p values shifted from 0.047 to 0.126, 0.002 to 0.005 and 0.037 to 0.044.When adjusting the burden of cSVD for vascular medical history, no meaningful changes were observed (all changes are < 10 %) (Supplementary Table 5).Finally, some dual-decliner groups were not solely of one domain.The number of participants with overlap in the dual decline in gait and memory domain was as follows (gait and processing speed n=20; gait and verbal fluency n=13).Due to this overlap, we conducted an additional post hoc analysis removing the overlapping individuals from the gait and memory group.The strength of associations (RRR) remained similar; however, the confidence intervals were wider and results nonsignificant, most likely due to the smaller sample size (n=16).

Gait speed and processing speed-attention (Digit Symbol Coding test) decline
Table 3 shows the associations of baseline brain volumes and total burden of cSVD with group membership for gait speed and processing speed-attention.Lower baseline gray matter (RRR 0.83 per 10 ml; 95 % CI 0.70, 0.97; p = 0.019) and white matter volume (RRR 0.88 per 10 ml; 95 % CI 0.79, 0.99; p = 0.037) were associated with a subsequent higher risk of dual decline group membership in the fully adjusted models.There was a linear association across WMH quartiles, with a higher burden of WMHs associated with a higher risk of gait only decline (RRR 1.66 per ml; 95 % CI 1.17, 2.35; p = 0.005) (Supplementary Table 6).Adjusting the gray matter (RRR 0.83 per 10 ml; 95 % CI 0.70, 0.97; p = 0.022) and white matter volume (RRR 0.88 per 10 ml; 95 % CI 0.78, 0.99; p = 0.028) for WMH did not meaningfully change the strength of the association with dual decliners.
No statistically significant interactions between age and sex and the MRI measures were found.

Sensitivity analysis
Age, stroke, diabetes, ischemic heart disease and depression were found to be predictive of missingness in the final analysis sample.In weighted models, coefficients were almost identical to the corresponding unweighted ones (Supplementary Table 8).Importantly, the associations for the primary analysis remained: point estimates for the adjusted models for hippocampal volume, gray matter and white matter moved from 0.59 to 0.57, 0.83-0.85and 0.82-0.80respectively (all these changes are <10 %), and corresponding p values shifted from 0.047 to 0.033, 0.037 to 0.057 and 0.002 to 0.001.

Discussion
We investigated the associations between baseline MRI brain measures and subsequent dual decline in gait and cognition across several cognitive domains.Consistent with our hypothesis, we found that a wide range of brain structures, representing neurodegeneration and vascular disease, were associated with dual decline in gait and cognition -both with respect to memory and processing speed-attention.Lower hippocampal volume was only associated with dual decline in the gait and memory group.No MRI measures were associated with dual decline with respect to verbal fluency.These findings contribute valuable insights into the brain structures associated with different gait and cognitive decline groups.
We hypothesized that prior findings of a stronger association

Table 4
Associations between baseline brain volumes and total burden of cSVD and group membership for gait speed and verbal fluency (Controlled Oral Word Association Test -FAS).between dual decline in gait and memory and higher risk of dementia might be due to this combination capturing a wider range of underlying brain pathology than dual decline in gait and non-amnestic domains (Collyer et al., 2022).Supporting this, our study demonstrated that lower hippocampal volume (along with gray and white matter volume) was associated with dual decline in gait and memory, but not other gait and non-amnestic groups.This is in line with the knowledge that lower hippocampal volume is associated with decline in memory (Cherbuin et al., 2015;Den Heijer et al., 2010).However, we additionally carried out two further analyses adjusting the hippocampal volume model for gray matter, white matter, and vascular medical history.When adjusting the model for total gray matter volume there was only a small reduction in the strength of association, suggesting that associations were not just due to a relationship between total gray matter volume and dual decliners.However, upon adjusting for both gray matter and white matter volume, the previously observed association of hippocampal volume disappeared, suggesting white matter atrophy may be an antecedent or confounder to hippocampal volume (Metzler-Baddeley et al., 2019).
When adjusting the hippocampal model for vascular medical history the strength of the association also reduced slightly.This suggests that these factors could also potentially serve as antecedents or confounders.As we are unable to partition these causal effects, and we are focused on capturing the overall burden of brain pathologies at baseline, irrespective of their origin, we presented these findings as post hoc analyses.
Interestingly there was an association between a higher burden of WMH and dual decline in gait and memory, but not the total cSVD score.This association might be explained by the effect of WMHs on cortical and sub-cortical networks that control both cognition and gait (Lu et al., 2021;Snijders et al., 2007), whereas the non-significant association with the total cSVD score may be due to the less sensitive way WMH are measured in the cSVD score (one point rating score).
Few prior studies have examined the associations between brain structure and dual decline in gait and cognition.One prior study demonstrated that greater WMH, lower total brain and hippocampal volume was associated with dual decline in gait and the Mini-Mental State Examination, a screening test of global cognition (Grande et al., 2023).Our results raise the possibility that the associations found in this prior study between lower hippocampal volume and dual decliners may have been driven specifically by the memory component of the Mini-Mental State Examination.In contrast to our findings, another study reported that there was no significant association between baseline hippocampal volume and dual decline in gait and memory (Tian et al., 2021).This may be due to the prior study using an immediate recall test, whereas we used delayed memory recall which is more specific to hippocampal pathology (Den Heijer et al., 2010).Interestingly, a recent study found a higher likelihood of dual decliners in gait and memory being carriers of Apolipoprotein E4 gene (Tian et al., 2021).The Apolipoprotein E4 gene is known to increase the risk of Alzheimer's disease, which is characterized by memory decline (Liu et al., 2013;Raulin et al., 2022).Taken together, findings suggest that a combination of vascular (WMH) and neurodegenerative pathologies are associated with dual decline in gait and memory.Incorporating blood biomarkers such as Aβ42/40 or amyloid PET imaging into future studies will provide a holistic understanding of the underlying brain pathology of dual decline in gait and cognition and the increased risk of dementia.
Our findings that gray matter and WMH volume were associated with increased risk of dual decline in gait and processing speed-attention is consistent with prior knowledge that lower gray matter and higher burden of WMH are associated with both slower gait (Callisaya et al., 2014a;Hairu et al., 2021) and processing speed-attention (Hong et al., 2015) individually.This association may be attributed to the compound effect of slow information processing and transmission through neural pathways, consequently contributing to the dual decline.Interestingly there was no evidence of an association between MRI measures and dual decline in gait and verbal fluency.This could be attributed to the relatively smaller number of participants (n = 32) and wider confidence intervals in the dual decline group.
This study has several strengths.Our sample was a population-based sample which was selected randomly from the electoral roll, making our findings more generalizable to the community.While previous studies were limited to a single cognitive domain, we were able to examine multiple cognitive domains.In addition, we employed a broader range of neuroimaging markers.However, there were also some limitations.While our study involved a population-based sample, it is important to note that our sample consisted of individuals of white ethnicity, due to the composition of the older population in Tasmania.Therefore, the generalizability of our findings to more diverse populations may be limited.The sample size in the dual decline groups across all the cognitive domains were small, and the follow-up time was short potentially reducing our ability to detect associations.We exclusively examined the impact of a limited number of vascular and neurodegenerative pathologies in our study.It is important to acknowledge that unobserved neurodegenerative pathologies such as beta-amyloid and tau accumulation could also contribute to dual decline in gait and cognition.Due to scanner changes halfway through the study, we were unable to reliably conduct longitudinal analyses on MRI variables.Additionally, the use of 1.5 Tesla MRI scanner may have introduced some limitations in terms of image resolution.Future studies employing higher field strengths and advanced imaging techniques could provide enhanced visualization and detailed assessments.Furthermore, we didn't look at specific locations of cSVD, gray matter (apart from hippocampal volume) or white matter, which may have associations with the dual decline in gait and cognition.Prior studies have found associations with specific locations of cSVD with gait (e.g., infarcts in the frontal lobe with lower gait speed) (De Laat et al., 2010) and specific gray matter locations with cognition (e.g., dorsolateral prefrontal cortex with verbal fluency) (Ghanavati et al., 2019).Finally, while we employed multiple cognitive tests to assess several cognitive domains (memory, processing speed-attention, verbal fluency) it is important to acknowledge that there were some overlaps among the dual decline groups.
This study builds on prior studies by showing the associations of both markers of neurodegeneration and cSVD with dual decline in gait and cognition across several different cognitive domains.Our findings demonstrated that a broad range of MRI measures were associated with dual decline in gait and memory.Taken together with prior studies, our findings suggest that the combination of decline in memory and gait points towards a mixed vascular and Alzheimer's type dementia, whereas dual decline in gait and other domains may be more specific to vascular dementia.However, further research examining specific Alzheimer's disease biomarkers such as amyloid PET brain scans and blood or CSF biomarkers of amyloid or phosphorylated tau proteins with dual decline groups is required.

23/04/2024
We confirm that this work described has not been published previously and, that it is not under consideration for publication elsewhere, that its publication is approved by all authors and tacitly or explicitly by the responsible authorities where the work was carried out, and that, if accepted, it will not be published elsewhere in the same form, in English or in any other language, including electronically without the written consent of the copyright-holder.
A/Prof Michele Callisaya University of Tasmania

Table 1
Participant characteristics for groups based on gait speed and memory decline (Hopkins Verbal Learning Test-Revised).

Table 2
Associations between baseline brain volumes and total burden of cSVD and group membership for gait speed and memory (Hopkins Verbal Learning Test-Revised).
Model 1 adjusted for age, sex, education level with the addition of total intracranial volume for volume measures.cSVD, cerebral small vessel disease; ml, milliliter; RRR, relative risk ratio; CI, confidence interval Interpreting the Relative Risk Ratio (RRR): the RRR values in the table show the risk of belonging to the decliner group in that column relative to the non decliner group, for a one unit increase in the brain volume measure versus no increase.For the categorical cSVD burden exposure, the comparison is between the burden group relevant to that row, versus the lowest burden group