Profiting of LC-MRI, we were able to evaluate the time-dependent degeneration of LC occurring in patients belonging to the AD clinical continuum. Even though this phenomenon has been already demonstrated in neuropathological studies (Braak et al. 2011; Theofilas et al. 2017), only a few neuroimaging longitudinal ones have been performed on LC (Jacobs et al. 2023; Dahl et al. 2023). In 2023, Jacobs and colleagues explored the variation of LC over time in a population of autosomal dominant AD cases. They found that the decrease in LC signal was associated with tau accumulation in the precuneus, assessed through PET (Jacobs et al. 2023). In the same year, Dahl and colleagues found that in cognitively intact elderly individuals, the reduction in LC signal is associated with worsening memory performance (Dahl et al. 2023), aligning with previous cross-sectional LC-MRI studies (Dahl et al. 2019; Liu et al. 2020a).

It is worth noting that our study is the first LC-MRI follow-up analysis performed in late-onset AD, and it parallels the neuropathology findings reported by Theofilas and colleagues in 2017 (Theofilas et al. 2017). In that study, the authors investigated the degeneration of the LC across neurofibrillary pathology-related Braak stages (BB stage-Braak et al. 2011) in a cohort of AD patients and aged controls, using stereological analysis to precisely estimate LC neuronal number and thus providing very reliable data concerning the involvement of LC during both physiological and pathological ageing (Theofilas et al. 2017). Theofilas and colleagues reported progressive LC neuron loss and atrophy along BB stages, from I to VI (Theofilas et al. 2017). Moreover, they found that LC cell number reduction becomes statistically significant in BB mid-stages (III-IV), with the rate of cellular death becoming dramatic in advanced stages (V-VI) (Theofilas et al. 2017). Interestingly, our radiological findings match quite well the neuropathological ones just quoted, as we observed a global decrease of the LCCR parameter, which can be considered as a proxy of LC neuronal density (Keren et al. 2015). This was shown considering both the whole population and each diagnostic group separately, since we observed a significant LCCR reduction not only in MCI individuals, but also in ADD patients which are well known to be affected by more advanced AD pathology stages (Thal et al. 2002; Braak et al. 2011). Furthermore, Theofilas et al. showed that LC volumetric shrinkage and atrophy are related to BB staging as well, and follow a progressive reduction pattern, which is particularly evident in the brain of subjects affected by I-to-IV BB stages (Theofilas et al. 2017). Even in this case, our results are in line with this post-mortem evidence, since we found a significant decrease in values of the volumetric parameter LCVOX in MCI individuals, while in ADD patients we only observed a trend, which, however, did not survive the multiple comparison correction. These data may suggest that LC-MRI enables the estimation of progressive LC atrophy, with a trend slope that could be steeper in the initial stages of the disease compared to what occurring in demented patients.

Finally, Grinberg’s group study showed that LC does not suffer from a physiological and age-dependent alteration, but, rather, its degeneration is strictly related to the occurrence of AD pathology (Theofilas et al. 2017). If, on the one hand, their observation is crucial for hypothesizing a specific role of LC impairment in AD pathogenesis, on the other it highlights the most important of the limitations of the present study, which is represented by the lack of radiological follow-up data on cognitively intact healthy controls. In line with this, we could not possibly draw any inference on LC-MRI features during healthy aging. However, it should be noted that such a limitation does not hamper the analysis we performed in patients, which clearly shows a time-dependent LC degeneration and is based on baseline MRI assessments in which LC loss of integrity was already present (Galgani et al. 2023). Moreover, the available literature might allow to speculate on what might be observed in cognitively-spared ageing. Cross-sectional studies reported that in the absence of cognitive decline, LC-MRI signal does not vary among different age groups (Giorgi et al. 2021; Al Haddad et al. 2023). Other studies revealed an age-dependent reduction of the signal in the rostral part of the LC, which however was inevitably related to a decline in the memory performances and in global cognition (Liu et al. 2019, 2020a; Dahl et al. 2019). The above-cited recent longitudinal study (Dahl et al. 2023) showed that reduction of LC signal during follow-up was associated with an increased risk of developing memory impairment (Dahl et al. 2023). Altogether, these pieces of evidence might suggest that LC-MRI signal may not undergo any kind of alteration in neurodegenerative-free ageing, as also shown by neuropathological evidence (Theofilas et al. 2017), while its decrease might be the sign of an ongoing degenerative phenomenon.

The second limitation of our study is the lack of concomitant in vivo neuroimaging information on the BB stages of patients included (e.g., Tau protein tracers PET) and of amyloid biomarkers. This reduced the accuracy of the evaluation of AD pathology in our sample, preventing us from linearly comparing our results with neuropathological ones and from excluding the occurrence of AD-mimicking pathology, such as primary age-related tauopathy (PART) (Crary et al. 2014) or limbic-predominant age-related TDP-43 encephalopathy (LATE) (Nelson et al. 2019; Liu et al. 2020b). Concerning PART, post-mortem data seems to indicate a common path with AD, with a similar involvement of LC by tau pathology (Kaufman et al. 2018; Zhu et al. 2019). For LATE, as far as we know, no neuropathological studies on the involvement of the LC have been performed yet and thus we cannot rule out its possible influence. To account for this limitation, we provided a thorough clinical, neuropsychological, and neuroradiological characterization of the study population (Galgani et al. 2023), and we believe that this might represent a solid background for discussing and interpreting our results.

Furthermore, we chose to include only patients affected by typical AD or amnestic MCI individuals. These subjects are more likely to bear AD pathology, with a predominant involvement of the limbic structures (Braak and Braak 1991). As the link of AD with LC degeneration has been defined by several studies (Dahl et al. 2019; Van Egroo et al. 2023; Galgani et al. 2023; Bell et al. 2023), this allowed us to frame our data and their interpretation in a well-explored context from a neuroanatomical and pathological point of view. In atypical forms of AD or non-amnestic MCI subjects, drawing similar assumption would have been more challenging. To the best of our knowledge, only one LC-MRI study has been performed up to now in patients with atypical AD (Olivieri et al. 2019). Even though those authors found that the LC degeneration occurs to a similar extent both in typical and atypical AD (Olivieri et al. 2019), the specific reasons for which AD pathology might affect the LC projections to either frontotemporal or occipitoparietal cortices rather than the limbic ones have not been elucidated yet.

The last limitation we want to acknowledge is the low number of patients involved (N = 57); even though this was sufficient to allow us to disclose the effect of time on LC degeneration, it probably hindered the accurate assessment of the effects of other factors, e.g. the baseline diagnosis.