Cross-sectional autopsy studies have suggested that, in Alzheimer’s disease, neurofibrillary tangles first appear in the locus coeruleus (LC) of the brainstem. Now, a longitudinal imaging study lends support to that theory. In the April 25 Nature Aging, scientists led by Jorge Sepulcre at Yale University in New Haven, Connecticut, and Heidi Jacobs of Massachusetts General Hospital, Boston, reported that LC degeneration seen on MRI, a proxy for tangles there, precedes tau PET positivity in the medial temporal lobe three years later. Transcriptomics hinted that dysfunctional intracellular protein transport might explain susceptibility to tauopathy in the LC and the medial temporal lobe. LC degeneration also preceded cognitive decline.

  • On MRI, locus coeruleus degeneration precedes tangles in the cortex.
  • On autopsy, tangles appear in the LC first.
  • Regional transcriptomes hint at common susceptibility to tangle pathology.

“This longitudinal biomarker study adds yet another piece to the compelling narrative that this group, and others, have supported so well over the years, positioning LC pathology as an early, key player in AD progression,” wrote Tharick Pascoal of the University of Pittsburgh (comment below).

Lea Grinberg at the University of California, San Francisco, called this an important paper for the field. “This study provides strong evidence supporting tau pathology spread from the LC to the entorhinal cortex, which has significant diagnostic and pathogenic implications,” she wrote. David Weinshenker of Emory University in Atlanta expressed similar sentiments, albeit with a caveat. “[P]ropagation of tau from the LC to the MTL has not been shown directly in human tissue,” he noted (comments below).

According to Braak staging, neurofibrillary tangles progress from the brainstem to the entorhinal cortex, then to regions of the allocortex, and finally the neocortex (Braak and Braak, 1991). Other autopsy studies have found tau aggregates in the LC before any other brain area (Ehrenberg et al., 2017; reviewed by Mravec et al., 2014). Neurons in the LC project into the cortex, leading scientists to wonder if toxic tau in the LC could travel and seed other areas.

LC First. LC intensity on MRI at baseline tightly correlated with tau PET signals in the left (left) and right (right) medial temporal lobe three years later (red). The opposite—baseline MTL tau and LC intensity at follow-up (gray)—was weakly associated. [Courtesy of Bueichekú et al., Nature Aging, 2024.]

Cross-sectional imaging and autopsy studies by Jacobs and colleagues had suggested as much. They showed that loss of LC integrity on MRI, a proxy for tangles, correlated with tau PET load in the medial temporal lobe (MTL), which includes the entorhinal cortex and hippocampus (Sep 2021 newsJacobs et al., 2023). They used MRI because PET of the tiny LC is not feasible due to the limited resolution and because tau tracers bind off-target neuromelanin, which concentrates there. However, despite having the MRI proxy measure, longitudinal data showing tau spread from the LC to cortex, or vice versa, was lacking.

To address this, first author Elisenda Bueichekú analyzed structural MRI and tau PET scans taken up to three years apart from 77 people from the Harvard Aging Brain Study (HABS), one of only a few studies with longitudinal MRI scans of the LC. Participants ranged from 41 to 89 years old, 65 percent were women, and three had mild cognitive impairment.

Those who had poor LC integrity at baseline had more MTL tangles three years later than did people whose LC was intact (image above). Yet high baseline MTL tau did not correlate with worse LC degradation at follow-up, suggesting that the LC degeneration, and by proxy, neurofibrillary tangles there, came first.

Nicolai Franzmeier, Ludwig Maximilian University, Munich, called the study elegant. “I was surprised how clearly LC abnormality precedes MTL tau,” he told Alzforum. Still, he cautioned that flortaucipir, the PET tracer used in HABS, has high off-target binding in the hippocampus. “The results should be replicated with other tracers that better capture MTL tau, such as MK6240 or RO948,” he said.

From Waning LC to MTL Tangles. Sagittal (left), coronal (middle), and axial (right) views of a tau PET scan suggest that poor locus coeruleus integrity at baseline (not shown) correlated with tau deposition in the hippocampus (red dotted circles) and amygdala after three years. [Courtesy of Bueichekú et al., Nature Aging, 2024.]

To look for direct evidence that tau deposits in the LC first, Bueichekú and colleagues turned to neuropathology data from 160 Rush Memory and Aging Project (MAP) participants. Sixty-six were cognitively healthy and 94 had MCI or AD. Their chances of having a higher-than-average tangle load in their LC but not in their hippocampus were 20 percent. Their chances of having unusually high tangle load in their hippocampus but not in their LC were only 0.05 percent. This supports the notion that tau deposits in the LC before the cortex, the authors concluded.

Did amyloid play a role in tau accumulation? Indeed, it may hasten its spread. Of 72 HABS participants who had baseline amyloid PET scans, 10 were amyloid-positive, i.e., above 18.94 centiloids. Notably, those clocking as little as 10 centiloids at baseline had more tau accumulation in their hippocampi, amygdalas, and fusiform cortices at follow-up if they had low LC integrity at baseline. “Even under only slightly elevated amyloid levels, the association between LC integrity and tau occurs in regions outside the MTL, which is consistent with the idea that amyloid facilitates further tau spreading,” Jacobs said. She envisions LC integrity as a very early marker of AD because it can detect tau accumulation earlier than amyloid PET or blood biomarkers.

LC degradation could also be a marker of imminent cognitive decline. While baseline MTL tau PET did not correlate with performance on the PACC5 three years later, MTL tau accumulation at follow-up due to weak LC integrity at baseline correlated with worsening PACC5 scores. This suggests that poor LC integrity predicted cognitive decline.

Why might tau accumulate in the LC, then spread to other regions? Curious if they share genomic susceptibility to tangles, Bueichekú analyzed bulk spatial transcriptomic data on healthy tissue of the LC, hippocampus, amygdala, cingulate cortex, and medial orbitofrontal cortex from the Allen Human Brain Atlas (Hawrylycz et al., 2012). Within the most highly expressed genes, i.e., the top 5 percent, in each area, the scientists identified 298 genes commonly expressed. These included three AD risk genes: progranulin, the γ-secretase subunit APH1B, and a putative cell adhesion protein, EPDR1 (Apr 2022 news).

Of the 298 genes, 34 fell into pathways that regulate protein transport, while two might affect tauopathy directly—the microtubule-associated protein MAP1β derails protein transport when it binds Aβ, while CDC42 regulates the tau kinase glycogen synthase kinase 3β (Gevorkian et al., 2008).

All told, the work suggests a common genetic profile between the LC and MTL that might leave these areas vulnerable to tangles. Jacobs and Sepulcre think the transcriptomics data will be a jumping-off point for other researchers to pursue ways to intervene in very early tau accumulation.—Chelsea Weidman Burke


  1. Bueichekú and colleagues conducted a study using multiple imaging biomarkers, including a 3T MRI sequence dedicated to locus coeruleus (LC) quantification, as well as amyloid and tau PET.

    They showed that LC integrity predicted longitudinal tau PET accumulation. The main novelty is the link between LC integrity and tau propagation over time. I really liked that even with a voxel-based approach, they found LC-related tau PET propagation mostly confined within the medial temporal lobe, where accumulation would be expected to occur in the population. A limitation that the authors could not overcome is the difficulty in quantifying tau in the small LC due to the limited spatial resolution of PET and the inherent off-target signal of the tau tracer around the area.

    Interestingly, the presence of amyloid facilitated the association and their model led to cognitive decline, nicely integrating their results with well-established models of AD progression.

    Their inferential transcriptomic results in the postmortem cohort suggested potentially regulated proteins that could be targeted—to be confirmed in future mechanistic studies.

    In summary, this longitudinal biomarker study adds yet another piece to the compelling narrative that this group, and others, have supported so well over the years, positioning LC pathology as an early key player in AD progression, which corroborates initial postmortem work.

  2. This paper is very important for the field in several aspects. It builds logically from the beautiful work that Heidi Jacobs has been doing on the importance of the neuromodulatory subcortical system, with a special focus in the locus coeruleus in AD progression.

    As the paper stresses appropriately, there is sufficient and rigorous postmortem evidence that the LC accumulates AD tau before the entorhinal cortex (EC-Braak 1 area), and that the tau burden in LC grows with Braak staging progression as its volume decreases.

    Jacobs’ previous studies highlighted the progression of LC integrity loss associated with AD progression and clinical outcomes. This paper successfully addresses several critical questions that remained open:

    1. Direction of Tau Pathology Spread: Is the progression of tau pathology in the LC and the loss of LC integrity primarily driven by pathology within the LC itself, or is it a reaction to tau pathology progression in the EC? This study provides strong evidence supporting tau pathology spread from the LC to the EC, which has significant diagnostic and pathogenic implications.
    2. Impact of Cortical Amyloid Pathology: The study explores how cortical amyloid pathology contributes to the spread of tau in relation to the tau burden in the LC, suggesting an additive effect.
    3. The utility of LC MRI metrics as biomarkers: Although current biomarkers identify AD pathology when it reaches moderate brain levels, few can predict the rate of cognitive decline. This study shows that including LC MRI metrics in the group of core biomarkers for AD significantly enhances the predictability of both the rate and speed of decline, which is crucial for clinical practice and for assessing the efficacy of drugs in trials.
  3. The noradrenergic locus coeruleus (LC) is among the first regions in the brain to accumulate hyperphosphorylated tau pathology, prior to the medial temporal lobe (MTL) and the cortex, in the prodromal stages of Alzheimer’s disease (AD). Because the LC projects directly to these structures, it has been proposed that the LC transmits and seeds tau pathology in the forebrain. While there is some evidence for this in rodent models of AD, the extent to which it occurs in humans is not known (Weinshenker 2018). 

    Heidi Jacobs and colleagues have helped pioneer the use of magnetic resonance imaging (MRI) for assessing human LC integrity in vivo, particularly in the context of aging/AD. By combining this approach with in vivo tau positron emission tomography (PET) and postmortem immunohistochemical analysis, these groups previously showed strong cross-sectional associations between LC tangle density, a loss of LC integrity, the accumulation of tau pathology in the MTL, and cognitive decline (Jacobs et al. 2023). Importantly, consistent with the idea that tau pathology propagates from the LC to the forebrain, the decline of LC integrity preceded the detection of tau in allocortical and neocortical structures. A notable caveat of this interpretation was the cross-sectional nature of prior studies, which prevented the spatiotemporal tracking of pathology within individuals.

    In the current study, the Jacobs group sought to overcome this limitation by implementing a longitudinal design, combining two time points of LC MR imaging, tau PET imaging, and cognitive data of 77 well-characterized individuals followed for up to three years. The main finding was that worse LC integrity at baseline predicted greater MTL tau pathology at follow up, but the reverse was not true, suggesting that tau accumulation in the LC precedes MTL pathology. Using data from the Rush Memory and Aging Project they corroborated this interpretation by showing that the likelihood of having tau tangles in the LC but not hippocampus is 400 times greater than having tau tangles in the hippocampus but not LC. Moreover, Aβ moderated the relationship between LC integrity and MTL tau, even at levels below the established Aβ-positivity threshold. Finally, greater LC-related MTL tau burden was associated with accelerated cognitive decline at follow-up, but baseline MTL tau was not.

    This study represents a major advance for the AD field, as the longitudinal findings dramatically strengthen the evidence for the LC as origin of cognitive impairment-related MTL tau pathology at the earliest stages of AD. However, this interpretation should be considered with caution because the propagation of tau from the LC to the MTL has not been shown directly in human tissue. While current technology does not permit this kind of analysis in vivo, recent advances in stem cell biology, e.g., induced pluripotent stem cell-derived LC-like neurons, and three-dimensional brain organoid assembloids, make in vitro experiments possible using human neurons for the first time (Tao et al., 2023; Birey et al., 2017). Another caveat is that tau PET cannot be used to assess tangle load in the LC; the small size of the structure is not compatible with the current resolution of tau PET, and the PET ligand cross-reacts with neuromelanin, which is found in high concentrations in the LC and confounds the analysis. Technical advances in tau PET (better resolution, different ligands) will be necessary to resolve this issue. Treatments that prevent the emergence of tau pathology in the LC retard its propagation to forebrain structures, and/or maintain the integrity of LC-norepinephrine transmission are gaining traction as potential disease-modifying therapies for AD progression and warrant further investigation (Levey et al., 2021Ehrenberg et al., 2023). 

  4. The is another in a series of articles discussing the integrity of locus coeruleus norepinephrine projection neurons and cortical tau pathology and cognition in Alzheimer’s disease. The authors examined whether LC changes precede tau deposition in the medial temporal allocortex and whether specific genetic features underlie the selective vulnerability of the LC and tau.

    That LC tau pathology precedes and drives tau in the medial temporal lobe memory circuit was proposed several years ago. Here, the authors employ a newly developed MRI method to evaluate spatiotemporal patterns of LC integrity, in combination with tau imaging, among elderly adults with a CDR score of 0 and a MMSE of 25, who were considered to have intact cognition. Overall, the authors report that higher LC intensity was associated with and predicted tau deposition in the MTL over a three-year follow-up period.

    As with most imaging studies, the scans only outline a general region of cortical tau labeling and do not provide data on which cortical laminar and neuronal subtype(s) are affected by the disconnection of LC norepinephrine. Since it is well known that tau pathology correlates with cognitive decline in AD, it was not surprising that the authors found that that greater LC-related MTL tau burden associates with worse cognitive performance. Since the LC is a small cell population in the brainstem, consisting of a dorsal forebrain and a ventral spinal cord projection system, MRIs of the integrity or type and location of tau pathology would enhance the argument.

    Perhaps the most interesting aspect of this work is the determination of the similarity between LC gene expression and other human brain regions including the hippocampus, amygdala, and insula, all components of the MTL allocortex, the oldest areas of the human brain. This suggests that evolutionary pressures exerted during the expansion of the human brain played a role in the selective vulnerability of select neurons to tau pathogenesis. Moreover, gene ontology expression analysis found a common biological background associated with the regulation of protein transport, suggesting that transcriptomic profiles of protein transport regulation and protein folding likely play an important role in tau pathogenesis for the LC-MTL connectome that increases the risk of AD. This initial phase of the illumination of transcriptomic pathways underlying LC-related tauopathies, is a first step in developing novel drugs targeting a specific brainstem to cortical tau connectome involved in the onset of AD.

  5. It is exciting to see the first paper investigating longitudinal locus coeruleus (LC) integrity and longitudinal tau-PET in vivo! Bueichekú and colleagues expanded on their previous pioneering studies. In a cohort of almost all cognitively unimpaired older adults, reduced LC integrity associated with follow-up measures of tau tangles in the medial temporal lobe, while associations were weaker for the reverse association (tau at baseline being related to LC integrity at follow-up). While baseline LC integrity was associated with subsequent tau levels, I wondered if the rate of change of LC integrity might also predict rate of change of tau tangles in the cortex. Further, in postmortem data, the regions showing the strongest associations with tau tangles in the LC were in medial temporal lobe; this was especially the case in unimpaired participants, providing complementary evidence that the LC-medial temporal associations are potentially crucial in early stages of tau spreading.

    I think the effect of Aβ is also important to note. Here amyloid facilitated LC-related tau at follow-up in temporal regions beyond the medial temporal lobe. While this result is perhaps not surprising, given the strong effect of amyloid on the presence of tau tangles, the authors report that this facilitation might occur early on, at levels lower than the typical threshold for elevated Aβ in this cohort. Interestingly, the brain regions showing the highest transcriptomic similarity to the LC included the hippocampus and the amygdala, as well as frontal regions where amyloid plaques deposit early in AD. These results suggest a related (and complex) interplay between the different regions and vulnerability to AD.

    As carefully stated by the authors, longer follow-up, and extending investigations to cognitively impaired populations, will help clarify the spatiotemporal changes linking LC integrity and accumulation of AD pathology.

  6. It is well known that certain subcortical modulatory nuclei, including the pigmented pontine nucleus locus coeruleus (LC), exhibit very early tau cytoskeletal pathology. According to earlier large-scale and detailed neuropathological studies by Heiko Braak and colleagues, LC tau pathology precedes cortical lesions, and the LC may be an early starting point of pathological tau spreading toward the trans-entorhinal cortex. However, there is still no clear, in vivo, longitudinal evidence for this.

    Exploring this problem in vivo is cumbersome, mainly due to the off-target binding of the currently used PET radioligands to neuromelanin pigment, and to the still limited spatial resolution of PET cameras, especially considering the small size and elongated shape of the human LC. At the same time, research on LC in the context of neurodegeneration is important, not only because LC is a potential epicenter of tau spreading towards the forebrain. This tiny nucleus is, in fact, a key hub in the CNS connectome: It is the major source of noradrenaline throughout the brain, and it is involved in multiple functions, such as sleep/wake and mood regulation, attention, learning, as well as various vegetative regulations. Accordingly, LC dysfunctions due to progressive intracellular tau pathology and consequent neurodegeneration contribute to substantial functional deficits, which overall may exacerbate virtually all cognitive, neuropsychiatric, and vegetative symptoms of the Alzheimer’s disease spectrum. In this paper Jacobs and colleagues have investigated the spatial-temporal relation of LC integrity and cortical tau pathology, applying the combination of in vivo neuroimaging, cognitive tests, and transcriptomics.

    This is the continuation of earlier elegant work from Jacobs, demonstrating that MRI-measures of LC intensity are associated with initial tau accumulation in the entorhinal cortex and with retrospective memory decline (Jacobs et al., 2021). Now her team extend their methodical repertoire, applying complex statistical approaches to explore the spatiotemporal pattern of LC integrity in the context of cortical tau accumulation and cognition. They also used a publicly available, adult human brain transcriptome database to reveal the specific biological characteristics of LC-related tau-accumulation in the human brain.

    As a core finding, Bueichekú et al. have demonstrated that reduced LC integrity, measured by 3T MRI, precedes tau accumulation in the medial temporal lobe, involving the hippocampus and parahippocampal structures, and that this hypothesized pathway of tau spreading is related to lower cognitive performance approximately three years later. Furthermore, they observed highly similar protein-coding gene-expression levels between the LC and limbic system structures, including the hippocampus, amygdala, rostral anterior cingulate cortex, medial orbitofrontal cortex and insula, and exploration of gene ontology identified “protein transport regulation” and “protein folding” as factors contributing to tau vulnerability. Importantly, these transcriptomics data were retrieved from individuals up till Braak NFT stage III, and further studies are needed to extend this approach to developed AD.

    Application of advanced statistical approaches on in vivo imaging and cognitive test data revealed further interesting details. For instance, in cases with Braak stage II and lower, the binomial exact test results indicated that the probability of having elevated tau tangles in LC but not the hippocampus was 20 percent, the likelihood of having high tau tangles in the hippocampus but not in LC was only 0.05 percent, while the probability of having tau tangles in both regions was 76 percent, further supporting the hypothesis of tau spreading from the LC to the allocortical areas. In addition, co-examination of LC integrity (3T MRI), cortical tau PET, and cortical Aβ PET (PiB PET) revealed that neurodegenerative processes in the LC impact future tau accumulation in medial temporal cortical regions, and that LC-related tau accumulation progresses to lateral tempo-occipital regions when Aβ is elevated here. Spreading outside the medial temporal cortex starts at subthreshold Aβ values, indicating that this LC–medial temporal lobe pathway of pathologic change occurs early in the disease cascade, and that spreading to neocortical regions is further facilitated by increasing levels of Aβ.

    All these results are not unexpected, but they elegantly support the pivotal role of early LC pathology in the spreading of pathological tau and in the cognitive decline noticed in the Alzheimer’s disease spectrum, again, using in vivo imaging data.

    As a limitation, it is important to note that LC integrity, measured by MRI, does not directly measure LC pathological tau accumulation, and its exact biological interpretation is still under investigation. However, earlier clinical-pathological studies from the same group showed a strong correlation between LC integrity and LC tau deposition (Jacobs et al., 2021). Still, further detailed neuropathology and in vivo imaging correlations may be needed to dissect the contribution of pathological tau, and perhaps other potential biological sources, to the LC MRI signal.

    In a broader context, as the authors also emphasize in the abstract, this study clearly shows that LC intensity can be a promising indicator (biomarker) for identifying the time window when individuals are at risk of disease progression. Overall, this paper is a very fine example of the integration of in vivo imaging, neuropsychological tests, neuropathology, as well as omics technologies in a cutting-edge study.


    . In vivo and neuropathology data support locus coeruleus integrity as indicator of Alzheimer's disease pathology and cognitive decline. Sci Transl Med. 2021 Sep 22;13(612):eabj2511. PubMed.

  7. This study is really exciting because it used longitudinal neuroimaging to understand factors that lead to tau accumulation in the brain and to later cognitive decline. We already knew that structural integrity of the locus coeruleus, a brainstem region of the brain that’s important for the production of the neurotransmitter, norepinephrine, has been associated with tau pathology burden at different snapshots of time during Alzheimer’s disease progression. However, the takeaway from this work is that disruptions in the structural integrity of the locus coeruleus precede Alzheimer’s-related tau accumulation and that this pathway of tau spreading associates with later cognitive impairment. This gives us insight that detectible disruption of the locus coeruleus is happening even before there are substantial levels of tau tangles in the brain. If we think of Alzheimer’s disease progression as a story, prior to this study, we only knew about the individual chapters where the locus coeruleus was playing a role, but this study is helping us put those chapters in order. This is important for future intervention research because it demonstrates that if we can detect these disruptions to the locus coeruleus before there is substantial tau burden, we can potentially intervene before this pathology ultimately impacts someone’s memories.

    For their exploratory analyses, it was really interesting to see that the locus coeruleus had similar gene-expression profiles to not only the medial temporal regions but also limbic areas, because this may tell us more about how tau pathology affects behavior. These limbic regions are important for emotional arousal and regulation, and early genetic changes in these regions might be associated with different psychopathology characteristics such as increasing depression or anxiety. This exploratory finding may suggest we take a closer look at these earlier emotional changes that happen before substantial cognitive and pathological changes. Overall, I’m excited about how this study is helping us understand the story behind Alzheimer’s disease progression.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Is a Waning Locus Coeruleus an Early Sign of Alzheimer’s Disease?
  2. Paper Alert: Massive GWAS Meta-Analysis Published

Paper Citations

  1. . Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol. 1991;82(4):239-59. PubMed.
  2. . Quantifying the accretion of hyperphosphorylated tau in the locus coeruleus and dorsal raphe nucleus: the pathological building blocks of early Alzheimer's disease. Neuropathol Appl Neurobiol. 2017 Aug;43(5):393-408. Epub 2017 Mar 31 PubMed.
  3. . Locus (coeruleus) minoris resistentiae in pathogenesis of Alzheimer's disease. Curr Alzheimer Res. 2014;11(10):992-1001. PubMed.
  4. . Waning locus coeruleus integrity precedes cortical tau accrual in preclinical autosomal dominant Alzheimer's disease. Alzheimers Dement. 2023 Jan;19(1):169-180. Epub 2022 Mar 17 PubMed.
  5. . An anatomically comprehensive atlas of the adult human brain transcriptome. Nature. 2012 Sep 20;489(7416):391-9. PubMed.
  6. . Amyloid-beta peptide binds to microtubule-associated protein 1B (MAP1B). Neurochem Int. 2008 May;52(6):1030-6. PubMed.

External Citations

  1. Allen Human Brain Atlas

Further Reading

No Available Further Reading

Primary Papers

  1. . Spatiotemporal patterns of locus coeruleus integrity predict cortical tau and cognition. Nat Aging. 2024 May;4(5):625-637. Epub 2024 Apr 25 PubMed.