Dementia is by WHO declared a public health priority (World Health Organization, 2012; World Health Organization, 2017). Globally, it is estimated that there were 57.4 million persons living with dementia in 2019, and this number is projected to increase to 152 million in 2050 (Nichols et al., 2022). Alzheimer’s disease (AD) is the most common cause of dementia, accounting for 57% (GjOra et al., 2021) to 80% (Alzheimer’s Association, 2022a) of all dementia cases. Although there is no curative treatment to date, a timely and accurate diagnosis is important for ruling out potential reversible conditions, provision of information and counseling, facilitating access to services and optimizing care, support of lifestyle changes that may delay progression, and possibility of participation in clinical trials (Liss et al., 2021).

Traditionally, a clinical diagnosis of dementia and AD is based on thorough anamnestic interviews with the patient and an informant, cognitive tests, assessment of neurological, physical, and psychiatric symptoms, supported by brain imaging (Medbøen et al., 2022). However, neuropathological changes occur up to decades before clinical manifestation of cognitive and functional impairments (Jack and Holtzman, 2013). With the current perspective of AD and other dementias as conditions with long-term pathophysiological changes occurring in clinical asymptomatic individuals, the use of biomarkers representing underlying pathology has been established for diagnostic purposes. The first detectable pathology of AD in the cerebrospinal fluid (CSF) is reduced concentration of amyloid-β 42 (Aβ₄₂), followed by an increase in phosphorylated tau (P-tau) and total-tau (T-tau; Palmqvist et al., 2019). Lower concentration of Aβ₄₂ in CSF represents abnormal aggregation of Aβ deposits in the brain, whereas increased P-tau signifies formation of neurofibrillary tangles in the neuron. These are considered highly specific for AD, while increased T-tau is a marker for neurodegeneration (Palmqvist et al., 2019). Although positron emission tomography (PET) imaging with tracers is now used to detect concentrations of Aβ and tau in the brain (Maschio and Ni, 2022), both PET and CSF biomarkers are unavailable in many settings (Frisoni et al., 2017). In a global perspective, there is an urgent need for simple tools that can help to improve the etiological of diagnosis of dementia.

Mobility, in terms of gait and balance, entails complex cognitive and motor processes, engaging several parts of the central nervous system (Valkanova and Ebmeier, 2017). Meta-analyses have concluded that poor gait performance predicts development of cognitive impairment and dementia (Beauchet et al., 2016; Sekhon et al., 2019). Other aspects of mobility, such as balance and dual-task cost, are also strongly associated with cognitive function, in particular with executive function (Muir et al., 2012; Tangen et al., 2014). The dual-task cost is the effect a cognitive task (e.g., talking or counting) has on gait (Montero-Odasso et al., 2009). To enhance the understanding of the specific nature of the relationship between mobility and cognitive decline, exploration of associations with underlying pathology is needed.

Studies have investigated the relationship between mobility and CSF AD core biomarkers in people with cognitive impairment and dementia. In a pilot study of 17 patients with mild AD, higher gait variability was associated with reduced Aβ₄₂/Aβ₄₀ and Aβ₄₂/Aβ₃₈ ratios (Koychev et al., 2018). In two larger studies of memory clinic patients, worse performance on dual-task gait tasks was associated with presence of tau-pathology, but not with Aβ-pathology (Ahman et al., 2019; Muurling et al., 2020). In the largest study to date (n = 299), including cognitively unimpaired older adults and memory clinic patients with subjective cognitive decline (SCD) or mild cognitive impairment (MCI), Nilsson et al. found that while a pathologic Aβ₄₂/Aβ₄₀ was associated with impaired balance and worse dual-task performance in the total sample, these associations were not statistically significant in the MCI group alone (n = 99; Nilsson et al., 2021). P-tau₁₈₁ was associated with dual-task cost in the total sample (n = 240) as well as in the MCI group (Nilsson et al., 2021). So far, the results are not conclusive, and more studies are needed to elucidate such as which measure of mobility outcome is closest related to the CSF AD core biomarkers.

The overall aim of this study was to explore the associations between four different aspects of mobility and AD core biomarkers in CSF in memory clinic patients with all-cause MCI and dementia, as well as in a subsample with MCI and dementia due to AD.

Descriptive data for the total sample are presented in Table 1, as well as comparisons according to severity of cognitive impairment. Only two patients had SCD; therefore, we have treated SCD and MCI (n = 13) as one group. The dementia group consisted of 46 patients with AD or mixed AD, two patients with DLB, and one patient with PPA. The dementia group had worse performance on the Mini-BESTest (p = 0.021) and worse TUG DTC (p < 0.001) than the SCD/MCI group.

There were no significant differences between the two cohorts regarding demographic characteristics, MMSE, or CSF biomarkers (Table 2). However, cohort 1 had significantly higher gait speed than cohort 2 (1.08 vs. 0.98 m/s, p = 0.020), and worse TUG DTC performance (−31.5% vs. −15.5%, p = 0.035).

Aβ₄₂ was significantly correlated with each of the mobility outcomes, with Spearman’s rank correlation coefficients ranging from 0.28 (gait speed) to 0.39 (Mini-BESTest). T-tau was significantly associated with TUG DTC (r_s −0.37; Table 3).

Results from the adjusted multiple regression analyses are presented in Table 4. Aβ₄₂ was significantly associated with Mini-BESTest (0.004 [0.002, 0.008], p = 0.040) and TUG (−0.003 [−0.006, <−0.001], p = 0.041) in the fully adjusted model and with TUG DTC (0.038 [0.009, 0.067], p = 0.012) in step 1 of the adjusted model (adjusting for age, sex, and comorbidity). T-tau was only associated with TUG DTC (−0.032 [−0.057, −0.006], p = 0.016) in step 1 of the adjusted model. P-tau₁₈₁ was not associated with any of the mobility outcomes in any of the analyses. The adjusted explained variance in the regression models with significant associations ranged from 0.06 for the model with Aβ₄₂ and TUG to 0.26 for the model with Aβ₄₂ and Mini-BESTest.

In the sensitivity analyses with the subsample that had MCI or dementia due to AD (n = 54), Aβ₄₂ was associated with Mini-BESTest (0.007 [0.002, 0.011], p = 0.003) and TUG (−0.004 [−0.007, −0.001], p = 0.021), but not with TUG DTC nor with gait speed in the adjusted regression models. P-tau₁₈₁ and T-tau were not significantly associated with any of the mobility outcomes in this subsample.

The main findings from this study of memory clinic patients were that while Aβ₄₂ was associated with several aspects of mobility, tau-pathology was not associated with any of the mobility outcomes after adjusting for diagnostic status.

The strongest correlations in our study were observed between the Mini-BESTest, representing dynamic balance control, and levels of Aβ_42. This association was also significant in the subsample with AD pathology, as well as in the adjusted models. While balance is closely associated with degree of cognitive impairment (Tangen et al., 2014; Yan et al., 2022), few studies have examined its relationship with degree of neurodegeneration. In Nilsson et al.’s study, longer time to walk the figure-of-eight test (requiring dynamic balance during walking) was associated with pathological Aβ₄₂/Aβ₄₀ in memory clinic patients with SCD and MCI, thus their results are in line with ours (Nilsson et al., 2021). However, in their study, this association was not statistically significant in the SCD or MCI groups separately. Further support for the association between balance and amyloid pathology can be seen through studies of falls, where a very large proportion of cognitively normal hip fracture patients had abnormal Aβ₄₂/Aβ_40, P-tau, and T-tau levels (Oh et al., 2018), and Aβ₄₂ was associated with shorter time to first fall in older adults with preclinical AD (Stark et al., 2013).

Gait and TUG has been studied in relation to amyloid and tau in several previous studies, with conflicting results. In our study, higher gait speed was correlated with higher levels of Aβ₄₂, but this association was not significant in the adjusted models. This is overall in line with two studies of memory clinic patients where Muurling et al. found no association between gait and Aβ₄₂ levels (Muurling et al., 2020), while Koychev and colleagues reported that gait was associated with the Aβ₄₂/Aβ₄₀ and Aβ₄₂/Aβ₃₈ ratios, but not with Aβ₄₂ (Koychev et al., 2018). In our study, the lack of association could be explained by gait speed being a too simple outcome to detect subtle changes. However, this is not the case in the studies by Muurling et al. (2020) and Koychev et al. (2018), where they used sophisticated instrumented gait assessments deriving data on qualitative aspects of gait. The TUG test represents more than gait alone, as in addition to walking it also involves rising up, turning, and sitting down. This complexity might explain why in contrast to gait speed, TUG was significantly associated with Aβ₄₂ in all our analyses. However, none of the other studies found significant associations between TUG and amyloid in memory clinic patients (Nielsen et al., 2018; Ahman et al., 2019; Muurling et al., 2020; Nilsson et al., 2021).

Dual-task gait cost is suggested as brain stress test to detect mobility problems in older adults, as it combines walking with a cognitive attention-demanding task (Cullen et al., 2018). However, in our study, the associations between TUG DTC and Aβ₄₂ and T-tau were no longer significant when we also adjusted for diagnostic status. Overall, these results are in line with findings from Nilsson et al. (2021) and Nielsen et al. (2018) while Ahman et al. (2019) did not find any association between biomarkers and increased dual-task gait cost. Interestingly, in our study, the associations between TUG DTC, Aβ₄₂ and T-Tau were not significant in the subgroup analyses, which may indicate that the association was driven by participants with non-AD pathology.

The only significant association between tau and the different mobility outcomes in our study was between T-tau and TUG DTC. Dual-task gait cost was also associated with tau in the other studies (Nielsen et al., 2018; Muurling et al., 2020; Nilsson et al., 2021) except from Åhman’s study where only the cognitive cost of the dual task was associated with tau (Ahman et al., 2019).

A potential source for some of the heterogeneous findings across these studies could be inclusion of different samples regarding age or degree of cognitive impairment. In our study, all participants were recruited from a memory clinic and most had mild cognitive dysfunction, also the patients with dementia. Thus, we have a sample with more severe cognitive impairment than for example Nilsson who did not include patients with dementia (Nilsson et al., 2021) or Muurling et al. (2020) and Nielsen et al. (2018) who also included cognitively healthy older adults in addition to patients with MCI and dementia. Åhman’s study included participants comparable to the present study, including memory clinic patients with SCD, MCI, and dementia, but their sample’s MMSE score (median score 25 points) was lower than in our study (median score 27 points; Ahman et al., 2019). Age is also important as pathological levels of Aβ₄₂ and tau becomes more prevalent with age (Jack et al., 2017). The different cohorts in the mentioned studies appear to be of similar age as ours, with Nilsson’s sample being the oldest with a mean age of 71.8 years and Koychev’s sample the youngest with mean age of 67 years (Koychev et al., 2018; Nilsson et al., 2021).

Dual-task gait requires attentional resources and executive functions (Hobert et al., 2011) and dual-task gait assessment is reported to increase the sensitivity of gait analysis to differentiate between MCI and cognitively healthy older adults (Bahureksa et al., 2017). In our study, the associations between mobility and levels of Aβ₄₂ are supported by the significantly worse performance by the dementia group compared to the combined SCD/MCI group on both TUG DTC and Mini-BESTest. Less well-explored than gait and dual task, Mini-BESTest was the mobility outcome with the strongest correlation with levels of Aβ₄₂. Dynamic balance may emerge as a motor skill closely interrelated with cognitive function, and more studies should explore this association at an earlier stage of the AD continuum. Still, it is important to keep in mind that the correlation between mobility and levels of Aβ₄₂ was rather weak if one aims to consider mobility as a surrogate marker for ongoing dementia pathology.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by the Regional Committee for Medical and Health Research Ethics South East Norway. The patients/participants provided their written informed consent to participate in this study.

GT, KS, KT, and A-BK designed the study. GT analyzed the data and drafted the manuscript. All authors contributed to the article and approved the submitted version.

The Southern and Eastern Norway Regional Health Authority supported the study through funding for GT (grant number 2016099).

A-BK, KP, and PB-W report work with Roche drug trial BN29553 and Novo Nordisk drug trial NN6535-4730 trial. A-BK report work with Boehringer-Ingelheim drug trial 1346.0023.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.