Odor identification (OID) dysfunction, characterized by an impairment of the ability to correctly identify common odors, is a well-established early marker of cognitive decline (Dintica et al., 2019; Tian et al., 2023), mild cognitive impairment (MCI) (Roberts et al., 2016; Dong et al., 2023), and all-type dementia (Adams et al., 2018; Laukka et al., 2023) and Alzheimer’s disease (AD) dementia (Murphy, 2019). Though OID starts to decline in the fifth decade of life (Zhang and Wang, 2017), recent studies have shown that a significantly accelerated decline predicts progression to MCI (Guo et al., 2023; Pacyna et al., 2023), dementia (Pacyna et al., 2023), and the transition between them, even when the performance falls within the normal range (Guo et al., 2023).

Despite increasing interest in the topic, some questions remain regarding the mechanisms underlying OID decline in aging and the observed associations with dementia and cognitive impairment. Besides degeneration of the peripheral olfactory system, changes in the central nervous system have been hypothesized to play a crucial role in OID decline (Bothwell et al., 2023; Parvand and Rankin, 2020), especially since OID ability has a significant cognitive component (Jobin et al., 2023b; Larsson et al., 2016). In this context, brain atrophy has emerged as one of the potential leading contributing factors of OID decline in aging. Previous population-based studies of cognitively normal (CN) elderly using structural magnetic resonance imaging (sMRI) showed that lower gray matter (GM) volume (Devanand et al., 2010; Diez et al., 2024; Dintica et al., 2019; Growdon et al., 2015; Kamath et al., 2022; Segura et al., 2013; Seubert et al., 2020; Tan et al., 2022; Vassilaki et al., 2017), thickness (Growdon et al., 2015), and integrity (Shrestha et al., 2023), as well as a faster volume decrease (Tian et al., 2023) particularly in medial temporal lobe (MTL) structures (e.g., hippocampi, amygdalae, entorhinal cortex) were associated with poorer OID. Furthermore, a recent study reported an association between smaller GM volume in olfactory, frontoparietal, and temporal regions with accelerated OID decline (Pacyna et al., 2023). Another study examined the longitudinal association between GM measures and OID exploring hippocampal subfield volumes, reporting an association between faster hippocampal tail, subiculum, CA4, and dentate gyrus decreases and accelerated OID decline in older adults (Menelaou et al., 2022).

Generally, there is a consistent link between GM measures of the MTL and OID in aging—for a systematic review, see Bothwell et al., 2023. MTL atrophy may also affect memory processing (Bettio et al., 2017), potentially impacting a key cognitive component of OID ability. Beta-amyloid deposition may be one potential process linking brain atrophy and OID decline in aging, as suggested by meta-analytic evidence indicating an association between amyloid PET and poorer OID performance (Tu et al., 2020).

So far, research in this area has primarily focused on markers of neurodegeneration. Notably, previous studies have reported associations of olfactory dysfunctions with cerebrovascular (Murphy et al., 2002; Omori and Okutani, 2020; Palmquist et al., 2020) and cardiovascular diseases (Palmquist et al., 2020; Roh et al., 2021; Schubert et al., 2015; Seubert et al., 2017) and related risk factors (Dong et al., 2021; Palmquist et al., 2020; Roh et al., 2021; Siegel et al., 2019). In this regard, a previously proposed cerebrovascular mechanism hypothesizes that a reduced blood flow due to atherosclerosis of the supplying blood vessels could lead to hypoxia affecting key olfactory processing areas (Schubert et al., 2015; Siegel et al., 2019). In addition, it is noteworthy that cerebrovascular co-pathologies are also common in AD and Lewy body diseases [i.e., Parkinson’s disease (PD) and dementia with Lewy bodies] (Attems and Jellinger, 2014; Hijazi et al., 2022), which have a high prevalence of olfactory dysfunctions (Bagherieh et al., 2023; Silva et al., 2018). Regarding brain sMRI markers of microvascular lesions, previous population-based studies in CN older adults showed mixed cross-sectional findings on the link between white matter hyperintensities (WMH) burden and OID, pointing to a negative association (Dong et al., 2023) or no relation (Buchanan et al., 2020; Devanand et al., 2010).

Overall, the evidence supports a link between brain atrophy and OID dysfunction. However, previous studies generally lacked markers of cerebral microvascular pathology, limiting them to WMH when included. Moreover, most studies were either based on cross-sectional brain data or on a single time-point olfactory assessment. Thus, the longitudinal association of microvascular burden and brain atrophy accumulation with OID decline remains largely unexplored. The lack of studies including white matter (WM) lesions, which may be extensible to other types of microvascular lesions, and longitudinal approaches has been highlighted as two of the main research gaps in the field (Bothwell et al., 2023). A deeper understanding of brain-olfaction relations is pivotal to determining which early brain changes are potentially related to olfactory deterioration in aging and predementia stages.

To advance knowledge about the brain correlates of olfactory deficits in aging, we aimed to examine whether (1) various sMRI markers of brain atrophy and microvascular lesions and (2) their progression are associated with OID decline in a population-based cohort of older adults.

Descriptive information for the study population can be found in Table 1. The mean age at baseline was 70.2 years (SD = 8.7), 59.9% were females, and the average amount of formal schooling was 12.7 years (SD = 4.3). When comparing participants with and without follow-up sMRI markers, those without follow-up markers were older at baseline, had fewer years of formal schooling, showed lower semantic memory performance, and displayed higher WMH and lateral ventricular volumes, along with lower hippocampal and total GM volumes (p < 0.05, Table 1). Importantly, the two groups did not differ significantly in OID, global cognition, global PVS count, proportion of current smokers and APOE ε4 carriers, having hypertension and diabetes mellitus, and presence of lacunes (p > 0.05, Table 1). In addition, participants without follow-up markers displayed lower performance in episodic memory, attention and executive function, language, and perceptual speed tests (p < 0.05, Supplementary Table 1).

The OID score decreased steadily during the follow-up period (p < 0.001, Figure 1). During the 15-year follow-up period, the average follow-up time for olfactory assessment was 10.4 years (SD = 3.6). The mean number of olfactory assessments per person was 2.9 (SD = 1.2). The AAC in the OID score was −0.17 items (95% CI, −0.21 to −0.14; p < 0.001), in a model controlling for sex, age, and education.

In this population-based cohort study, we investigated associations of sMRI markers suggesting microvascular lesions and neurodegeneration with OID decline across 15 years. We found that (1) higher PVS count and lower hippocampal and total GM volumes at baseline, and (2) occurrence of lacunes over follow-up, faster increase of WMH volume, global PVS count, and lateral ventricular enlargement, hippocampal, amygdalar, and total GM atrophy were associated with accelerated OID decline. While previous studies mainly pointed to the involvement of brain atrophy process in age-related olfactory decline, our results added to the current literature by suggesting the contribution of cerebral microvascular lesions to an accelerated OID decline.

Our findings on the relationship between markers of cerebral microvascular lesions and OID decline provide new insights into the potential mechanisms underlying olfactory deficits. We found that a higher PVS count at baseline and its increase over time are associated with an accelerated OID decline. PVS are fluid-filled spaces surrounding cerebral small vessels, which are involved in the clearance of soluble waste products (Gouveia-Freitas and Bastos-Leite, 2021). It has been proposed that a reduced outflow through arterioles due to microvascular changes (e.g., atherosclerosis, arteriolosclerosis, elastin dysfunction) may alter their functioning, resulting in the accumulation of waste products and their enlargement (Gouveia-Freitas and Bastos-Leite, 2021). Recent neuropathological findings suggest that visible (i.e., enlarged) PVS may reflect an impaired clearance of beta-amyloid protein, contributing to its accumulation in the underlying cerebral cortex and promoting cerebral amyloid angiopathy (Perosa et al., 2022). Thus, the observed associations of OID decline with PVS and their increased count over time may reflect both microvascular burden and AD-related protein deposition in key olfactory areas in aging. In addition, it is suggested that PVS enlargement might occur before the accumulation of other microvascular lesions and neurodegeneration (Francis et al., 2019; Li et al., 2022), and it has emerged as an early predictor of progression to dementia (Paradise et al., 2021; Romero et al., 2022). Considering this perspective, the decline in olfactory function might hold significance as an early sign of structural brain aging.

To go further, we explored the regional counts of PVS and found that their count in basal ganglia and their accumulation in the frontal lobe are associated with accelerated OID decline. Visible PVS in the basal ganglia were linked previously to cerebral ischemia (Gouveia-Freitas and Bastos-Leite, 2021), which aligns with the hypothesized cerebrovascular mechanism of hypoperfusion causing vascular damage due to atherosclerosis of the supplying blood vessels (Schubert et al., 2015; Siegel et al., 2019). On the other hand, brain atrophy might also contribute to PVS enlargement by ex-vacuo dilatation secondary to shrinkage of cerebral tissue, even though there is mixed evidence and the underlying mechanism remains controversial (Zhang et al., 2016). In our study, the relationships between enlarged PVS and OID decline might be considered independent of brain atrophy since the results for baseline counts and progression remain significant after introducing all markers in the same model. Notably, in a sample of individuals with PD from the Parkinson’s Progression Marker Initiative (PPMI) cohort, a higher PVS count in the basal ganglia was associated with worse OID (Fang et al., 2020). Indeed, beyond the specific underlying processes, a presumable basal ganglia dysfunction associated with OID and its decline is coherent with evidence of their involvement in higher-order olfactory functions (i.e., odor encoding and recognition) (Eek et al., 2023). Similar mechanisms may be behind the association of OID with the accumulation of PVS in the frontal lobe, which encompasses some key olfactory regions (e.g., orbitofrontal and piriform cortices) structurally and functionally related to poorer olfactory functions in aging (Felix et al., 2020; Pellegrino et al., 2016; Shen et al., 2013; Seubert et al., 2020; Wu et al., 2019). Following this approach, future research should address whether different types of lesions beyond PVS may contribute to olfactory decline depending on the affected structures.

We also investigated if sleep quality at baseline could impact the relation between PVS and OID decline. After adjusting the analyses for self-reported reduced sleep, the results remained the same. Future studies should address the impact of objective sleep parameters, beyond the use of self-reported sleep measures, on the dysfunction of the glymphatic system, as a potential mechanism for olfactory deterioration in aging, as recently reported regarding memory decline (Ma et al., 2024).

In addition, the presence of lacunes at follow-up and the accelerated increase of WMH volume were associated with faster OID decline, aligning with the hypothesis of olfactory deficits as a consequence of vascular damage in key olfactory processing areas (Schubert et al., 2015; Siegel et al., 2019). This hypothesis is supported by evidence of associations of carotid intima-media thickness and atherosclerosis with OID decline in humans and middle cerebral artery occlusion causing olfactory cortex infarcts in rats (Duverger and MacKenzie, 1988; Schubert et al., 2015). In contrast, it is noteworthy that previous population-based studies have shown mixed results when analyzing possible cross-sectional relations between WMH and olfactory function (Devanand et al., 2010; Buchanan et al., 2020; Dong et al., 2023). In this regard, our results revealed no link between baseline WMH volume and OID decline, which aligns with those studies using cross-sectional designs that found no association (Devanand et al., 2010; Buchanan et al., 2020). The absence of a relationship between baseline WMH burden and OID decline contrasts with the observed longitudinal relationship. The divergence of results may be influenced by the large variability of WMH, OID, and their trajectories in aging, where, an acceleration may signal a more severe pathological process, such as an impending dementia disorder (Silbert et al., 2012; Guo et al., 2023; Pacyna et al., 2023).

Regarding neurodegeneration markers, we found an association between lower hippocampal volume at baseline and faster OID decline, complementing prior findings showing that OID is positively related to hippocampal volume and integrity in aging (Seubert et al., 2020; Shrestha et al., 2023; Diez et al., 2024), and that OID dysfunction is associated with lower hippocampal volume (Dintica et al., 2019) and faster hippocampal atrophy (Tian et al., 2023). Furthermore, lower total GM volume at baseline was associated with accelerated OID decline, which aligns with a widespread pattern of reduced GM volume found in faster decliners, including olfactory and other AD-related regions (Pacyna et al., 2023). This finding is also congruent with associations between GM measures in other regions beyond MTL and poorer OID in CN older individuals (Bothwell et al., 2023; Diez et al., 2024; Shrestha et al., 2023; Tian et al., 2023).

Another contribution of our study is that we investigated the relation between longitudinal brain atrophy trajectories and future OID decline. We found a relationship between faster hippocampal, amygdalar, and global atrophy and accelerated OID decline. Overall, this finding reveals that an early atrophic process, affecting specifically AD-related areas, is associated with olfactory deterioration in aging, as prior cross-sectional evidence suggested (Devanand et al., 2010; Dintica et al., 2019; Seubert et al., 2020; Pacyna et al., 2023; Shrestha et al., 2023; Tian et al., 2023). This result complements the 5-year associations of volume decreases in specific hippocampal subfields with OID decline previously reported (Menelaou et al., 2022). Relatedly, a recent study reported that poorer OID was associated cross-sectionally with lower GM matter volume and longitudinally with tau accumulation in key olfactory areas, including MTL structures in CN older adults (Diez et al., 2024). In addition, the authors found that tau spreading in the brain followed a pathway from MTL structures toward the olfactory system (Diez et al., 2024). Following this direction, future studies should investigate which early regional brain atrophy trajectories underlie OID decline. One promising path to address this question is exploring the longitudinal trajectories of core regions of the functional olfactory network, following previous cross-sectional studies exploring these regions in relation to OID and episodic memory performance (Seubert et al., 2020; Jobin et al., 2023a).

Interestingly, some associations did not remain significant when entering all sMRI markers in the same model. This finding may reflect a joint effect of cerebral microvascular lesions and neurodegeneration on olfactory decline. Especially, considering their interactive effects (Paolini Paoletti et al., 2021) and that most dementia cases in community-based cohorts and brain banks exhibit mixed dementia pathology (Schneider et al., 2007; Agrawal and Schneider, 2022; Robinson et al., 2023). The interpretation should be cautious, especially for sMRI markers suggesting microvascular lesions. For example, neuropathological evidence showed that WMH corresponds to areas of demyelination and gliosis (i.e., neuronal loss) (Young et al., 2008), and a recent study indicated that WMH contributes to ventricular enlargement (Jochems et al., 2022). Regarding enlarged PVS, as we highlighted above, their formation may be due to an initial composition alteration of the perivascular fluid (Gouveia-Freitas and Bastos-Leite, 2021) or be related to brain atrophy (Zhang et al., 2016). Future studies should address how these different processes could interact in their association with olfactory deterioration in aging and predementia stages through implementing mechanist models.

We detected a significant three-way interaction between sMRI markers, follow-up time, and age groups. Higher baseline PVS count was more strongly associated with accelerated OID decline in old-old adults, who have previously demonstrated a faster decline (Ekström et al., 2020). This interaction with baseline PVS may reflect the broader relationship between PVS accumulation and olfactory decline, suggesting that the observed age-specific effect is a more pronounced manifestation of this overall pattern. On the other hand, we found that lower hippocampal volume at baseline was more strongly associated with accelerated OID decline in males. It is common knowledge that sex impacts olfactory function, with females outperforming males (Sorokowski et al., 2019). Moreover, sex differences in brain olfactory regions have also been reported (Alotaibi et al., 2023), including lower GM concentration in the hippocampi in males (Garcia-Falgueras et al., 2006). Despite this, longitudinal analyses point out that hippocampal atrophy appears to impact olfactory deterioration in both sexes. Finally, lower hippocampal volume at baseline and faster GM atrophy in APOE ε4 carriers were associated with accelerated OID decline, which may reflect an ongoing pathological process and early functional changes in a group that already presents a high AD genetic risk (Calhoun-Haney and Murphy, 2005; Mishra et al., 2018). Indeed, APOE ε4 has been reported to be associated with smaller volumes of specific olfactory areas (i.e., amygdalae and entorhinal cortex) (Pacyna et al., 2023).

It is noteworthy that the main findings remained significant after controlling for current smoking and semantic memory performance. These results point to central nervous system mechanisms independent of smoking effects on the olfactory function (Ajmani et al., 2017). On the other hand, they reinforce the notion that memory has a significant but modest effect on OID (Jobin et al., 2023a), despite potentially shared mechanisms underlying their decline, such as MTL atrophy (Bettio et al., 2017; Bothwell et al., 2023).

In the sensitivity analyses excluding anosmic individuals, the global PVS count at baseline was no longer associated with OID decline, while the PVS increase remained significant. It is important to note that participants classified as anosmic still display olfactory decline, which did not significantly differ from the observed in those without anosmia. We further found that the associations of baseline brain sMRI markers with OID decline did not remain significant after excluding individuals diagnosed with dementia or PD during follow-up. Of note, those participants with incident syndromes had lower OID and cognitive scores and displayed a worse profile for microvascular lesions and neurodegeneration at baseline. Moreover, they showed a faster decline in their ability to correctly identify smells. Thus, these individuals were likely in an advanced early stage of neurodegeneration, characterized by ongoing structural and functional pathological processes, which might explain why removing them from the sample led to nonsignificant associations with the baseline sMRI markers. In contrast, the longitudinal associations for all markers remained significant after the exclusion of these participants, except for the presence of lacunes and amygdalar atrophy.

Previous research has mainly targeted olfactory performance as a marker of future brain atrophy and dementia incidence (Adams et al., 2018; Laukka et al., 2023; Pacyna et al., 2023; Tian et al., 2023). The current findings suggest that olfactory dysfunction potentially develops in parallel with changes related to both cerebrovascular burden and neurodegeneration, which occur early in aging and dementia development, some of them affecting key areas for high-order olfactory and cognitive functions (i.e., hippocampi, basal ganglia, and frontal lobe). In this regard, we collected new evidence supporting central nervous system changes as a potential mechanism of OID decline in aging (i.e., hippocampal atrophy) and propose new candidates (i.e., PVS enlargement in basal ganglia and frontal lobe) for which we may expect an overacceleration in the progression to dementia. Future studies should address the risk of developing olfactory dysfunction in aging based on MRI biomarkers, as it predicts future dementia (Adams et al., 2018; Murphy, 2019; Laukka et al., 2023).

The major strength of this study is the population-based design with a long follow-up period for brain sMRI markers (6 years) and OID (15 years), which allowed us to analyze their longitudinal associations. Significantly, we extended previous studies focusing on neurodegeneration by adding cerebrovascular burden markers. A possible limitation is the use of diverse automated and manual approaches for the assessment of sMRI markers, which could affect the replicability in different samples. Furthermore, the regional component was not extensively addressed for other markers apart from subcortical MTL regions and PVS.

The present study lacks protein biomarkers of AD pathology. Of interest, a population-based study in CN older adults linked faster accumulation of beta-amyloid in the orbitofrontal cortex and tau in the entorhinal cortex to accelerated OID decline over 2.4 years (Tian et al., 2022). Future studies should longitudinally explore the link between olfactory function, microvascular lesions, and neurodegeneration also including a range of available molecular markers. Considering multiple pathological processes may result in models bringing us a more detailed picture of which brain changes can lead to olfactory decline while outlining high-probability trajectories of conversion to dementia. The resulting models could be transferable to clinical settings through early detection approaches by combining olfactory assessment with, for example, blood-based markers (Xiao et al., 2023). Another constraint is the participants’ profile, which encompasses an older population residing in the central area of Stockholm, relativity healthy and fit, predominantly high socioeconomic standing, and Swedish-born. These characteristics might limit the generalizability of our findings.

Overall, in this population-based cohort study, we report a longitudinal association of the progression of microvascular lesions and neurodegeneration with OID deterioration in aging. Further research should model how the longitudinal interplay of different brain pathological processes may impact the sense of smell. These mechanistic models may increase our understanding of olfactory decline in aging and the period leading up to a dementia diagnosis, facilitating early dementia detection.