To enable clinical trials in frontotemporal dementia, scientists need biomarkers that distinguish its several underlying pathologies. At a recent conference in Miami, speakers showed data on potential fluid biomarkers and PET tracers that could differentiate the two main proteinopathies, tau and TDP-43. They also discussed nascent markers and debuted an AI algorithm that improves the diagnostic capabilities of FDG-PET scans.
Second Holloway Summit Showcases Intense Search for FTD Biomarkers
Because frontotemporal dementia is a rare disease, it has not gotten nearly the attention or research dollars of its big cousin, Alzheimer’s. As a result, progress in the field has lagged behind, particularly with regard to biomarkers. Philanthropist Kristin Holloway, whose husband, Lee, got a diagnosis of FTD at the age of 36, is trying to change that by funding an annual meeting to convene leaders in the field and discuss areas of need. Organized by the Association for Frontotemporal Degeneration, the first such meeting last year focused on digital biomarkers (Jun 2022 conference news). The second Holloway Summit, held November 30 to December 1 in Miami, turned the spotlight on biomarkers, of both the fluid and imaging kind.
Biomarkers are challenging in FTD because the disease is more heterogeneous than AD, with several different underlying pathologies. About 40 percent of patients accumulate tau fibrils, nearly half TDP-43 deposits, and the remainder have a variety of other proteinopathies, such as TMEM106B or indeed even TAF15 (Dec 2023 news). Over the two-day Holloway summit, some 50 academic and pharmaceutical scientists assessed the state of the field on biomarkers specific for pathologic tau and TDP-43, and also brainstormed ways to accelerate discovery. The conference was a discussion and networking opportunity more than an occasion for showing unpublished data.
Participants said the discussions would spur their own research. Arthur Simen of Takeda was intrigued by the potential use of FTD biomarkers in trials, noting, “This meeting has given me reason to be hopeful.” Henrik Zetterberg of Gothenburg University, Sweden, said that biomarkers that reflect core molecular features of FTD pathologies look promising. “This bodes well for the future,” he told Alzforum.
Mix and Match. FTD clinical syndromes (top)—such as behavioral variant FTD, semantic variant primary progressive aphasia, non-fluent variant PPA, FTD-motor neuron disease, corticobasal syndrome, and progressive supranuclear palsy-Richardon’s syndrome—can be caused by numerous underlying pathologies (bottom), including variants of tau, TDP-43, FUS, and UPS. [Courtesy of Grossman et al., Nature Reviews.]
Is It Tau or TDP-43?
Potential disease-modifying therapies are being developed for different forms of FTD. To recruit the right patients into trials, researchers need a quick way to determine what type of pathology the person in front of them has. About 40 percent of FTD patients have a family history of the disease, and specific mutations do correlate with particular deposits—C9ORF expansions and GRN mutations with TDP-43, MAPT mutations with tau. Alas, for sporadic cases, there is currently no way to tell what deposits are present. Complicating matters, tau fibrils can consist of either three- or four-repeat isoforms. Both 3R tauopathies such as Pick’s disease, and 4R ones such as progressive supranuclear palsy and corticobasal degeneration, fall under the FTD umbrella, and people with behavioral variant FTD can have either type of tau (see image above).
In the last year, new studies have proposed fluid biomarkers for both tau and TDP-43 deposits. In Miami, ChihiroSato, Washington University St. Louis, presented her work on two mid-domain tau phosphorylations, MTBR-275 and -282. Their concentration is elevated in the parenchyma, and decreased in the cerebrospinal fluid, of people with 4R tauopathies, much like Aβ42 in the Alzheimer’s disease brain. The CSF markers are particularly low in CBD, notching an AUC of 0.83 for distinguishing this disorder from other dementias, including from other 4R tauopathies (Dec 2022 news). Sato said she next will validate this assay in real-world clinical populations. She would also like to correlate the biomarker with tau PET, and move the assay to blood.
HDGFL2 Marker. A cryptic peptide is higher in the CSF of C9ORF72 expansion carriers and ALS patients (red) than in controls (black), even before FTD symptoms arise. [Courtesy of Philip Wong and Katherine Irwin.]
Meanwhile, Leonard Petrucelli at the Mayo Clinic in Jacksonville, Florida, and Philip Wong at Johns Hopkins University School of Medicine, Baltimore, discussed the presence of aberrant cryptic peptides in certain proteins as a potential new TDP-43 biomarker. TDP-43 prevents such splicing errors, so when TDP-43 exits the nucleus and deposits in the cytosol, more cryptic peptides arise. Some cryptic peptide-containing proteins, such as HDGFL2 and MYO18A, can indeed be detected in CSF (Feb 2023 news). In Miami, Wong added data from postmortem brains of ALS patients and C9ORF mutation carriers, showing that when neurons in motor cortex and hippocampus lose nuclear TDP-43, cryptic HDGFL2 appears.
CSF assays for cryptic HDGFL2 have now been done in 161 C9ORF72 expansion carriers, up from the previous 46, and continue to show good discrimination from controls, Wong said. In the 83 presymptomatic carriers tested so far, levels were about twice as high as in 66 controls, while in the 78 symptomatic carriers, they were three times as high.
The biomarker also works in sporadic disease, where 44 people with ALS notched levels double those of controls. Petrucelli noted that researchers are currently optimizing assays for both cryptic HDGFL2 and MYO18A in preparation for further testing and validation. Ideally, clinicians would prefer plasma assays to CSF. Wong said this might be possible, as the plasma signal for cryptic HDGFL2 correlates well with the CSF signal, at r=0.79.
Other potential candidates to detect TDP-43 pathology in fluids are specific for certain mutations, such as progranulin in GRN mutation carriers, and dipeptide repeat proteins in C9ORF expansion carriers. Scientists are exploring additional biomarkers. In Miami, Giacomo Salvadore of Alector introduced some of these, noting that symptomatic GRN mutation carriers have elevated lysosomal proteins, such as cathepsin D and LAMP1, and complement proteins such as C1q and C3b. At the moment, these markers can only be reliably measured in CSF, limiting their use.
Scott Small of Columbia University in New York noted that many FTD and ALS genes relate to late endosomal function, whereas Alzheimer’s genes tend to be found in the early endosome, and Parkinson’s genes in the lysosome. For example, C9ORF72 expansions block maturation of the early endosome, and CHMP2b is part of the endosomal sorting complex that helps recycle receptors. TMEM106B, which helps transport late endosomes, can form fibrils in FTD brain (Apr 2022 news).
Powering Discovery Efforts
In addition, the summit highlighted efforts to better validate existing analytes in body fluids. Consider neurofilament light. NfL is one of the most useful biomarkers for FTD. Though nonspecific, it rises in the CSF before symptoms appear, and tracks with clinical progression (Jun 2016 news; Sep 2016 conference news; Mar 2021 conference news).
Clinician-researchers would much prefer to measure NfL in blood rather than CSF, but NFL measurements there are notoriously variable. ALLFTD, the North American observational study of FTD mutation carriers, has started the Neurofilament Surveillance Project, in which nurses will visit the homes of about 335 mutation carriers every three months for three years to draw blood. These quarterly measurements will be compared with annual NfL draws done in a doctor’s office, as well as with numerous other health factors, to try to pin down sources of variability. The idea is to help ready plasma NfL as a biomarker for trials.
Like NfL, the inflammation marker GFAP is not disease-specific, rising in all symptomatic FTD cases and in other neurodegenerative diseases. On the up side, it too can be measured in blood. Numerous talks in Miami debated possible applications for these nonspecific markers. For example, a GFAP/NfL ratio would be higher in AD than FTD and might help distinguish the two diseases, noted David Irwin at the University of Pennsylvania, Philadelphia. However, in general these markers are more useful for tracking disease progression and severity than for diagnosis.
How about new markers? One such effort is the BEYONDD study led by Adam Boxer and Gil Rabinovici at the University of California San Francisco, as well as Desiree Byrd at CUNY Queens College and Monica Rivera-Mindt at Fordham University, New York City. The goal is to recruit more black and Latino people under 65 who develop dementia. Early onset cases are more likely to be FTD than AD, which tends to happen later in life. Participants can complete questionnaires and neuropsychology tests online, and have blood draws done at home. This may turn up plasma markers with broad applicability in real-world populations.
Similarly, Yolande Pijnenburg at VU University Medical Center is establishing a young-onset dementia cohort in the Netherlands. People can register at any point in their disease for the study, which will offer education and support, and will invite participants to donate blood samples. Recruitment will start in early 2024.
Meanwhile, Charlotte Teunissen at Amsterdam UMC is analyzing CSF protein panels to find markers that discriminate between FTD-tau and -TDP. So far, the project has identified panels that identify FTD from controls with AUC=0.96, and from AD patients with AUC=0.91, Teunissen said via video conference. Alas, an initial 10-protein panel that distinguished tau and TDP cases did not replicate.
Another marker of interest is acetylated tau, which has been linked to 3R tauopathies and AD (Sep 2015 news; Oct 2023 news). Because this marker is higher in FTD than AD, a CSF p-tau217/acetylated-tau ratio might discriminate the two diseases, Teunissen suggested. The hunt continues.
For data on imaging biomarkers, see Part 2 of this series.—Madolyn Bowman Rogers
At the second annual Holloway Summit, held November 30 to December 1 in Miami, frontotemporal dementia researchers showcased the current state of knowledge in finding biomarkers that identify different underlying pathologies of the disease. For broad-based, routine clinical applications, fluid biomarkers will be crucial. In this area, researchers now have candidates that could distinguish between tau and TDP-43 deposits (see Part 1 of this series).
Their other ask is to come up with reliable imaging biomarkers for research and some narrower clinical applications. Here too, attendees had recent developments to highlight. PET tracers for both pathologies are in clinical trials, while the old-fashioned FDG-PET scan has been given a boost into the 21st-century courtesy of machine learning algorithms. With this enhancement, such scans can differentiate multiple neurodegenerative diseases.
Even so, attendees agreed that more is needed to power FTD research. In particular, new biomarkers should be integrated into natural history studies to generate detailed biologic staging schemes. This would put FTD closer to the standard set by Alzheimer’s and Parkinson’s research, and enable disease-modifying trials.
Telltale Uptake. Different brain structures light up with tau tracer APN-1607 in CBD (left), PSP (middle), and Pick’s disease (right). [Courtesy of Tagai et al., Neuron.]
One Tau Tracer to Bind Them All?
Researchers have made progress on PET for each of the two main FTD proteopathies. In Miami, Brad Navia of Aprinoia Therapeutics showed data on the company’s tau tracer APN-1607. It is one of two tracers, along with Life Molecular Imaging’s PI-2620, which detects 4R fibrils, and its uptake correlates with disease severity in PSP patients (Mar 2020 conference news).
APN-1607 binds all types of tau fibrils, but it lights up distinct areas of the brain in different tauopathies, Navia noted. For example, in PSP the signal emanates from deep brain regions, whereas in AD it is most prominent in the temporal cortex. These patterns could help determine which pathology underlies a broad clinical syndrome such as bvFTD, Navia said. He showed APN-1607 scans of four bvFTD patients in whom one had more cortical uptake, suggesting a 3R tauopathy like Pick’s disease, and another had primarily subcortical uptake, suggesting a 4R tauopathy. A third had an AD-like pattern, and the fourth had no tracer uptake, potentially indicating the presence of another pathology such as TDP-43 (image below).
Tauopathy Patterns. The tau tracer APN-1607 helps distinguish underlying pathologies of behavioral variant FTD, showing patterns characteristic of 3R tau (upper left), 4R tau (lower left), or AD (lower right). Absence of a signal (upper right) could indicate other pathologies such as TDP-43. [Courtesy of Makoto Higuchi.]
Thus far, Aprinoia has scanned about 400 people with PSP and 50 MAPT mutation carriers, most of whom are still living and being followed. APN-1607 distinguishes PSP cases from controls with 94 and 96 percent sensitivity and specificity, respectively, Navia said. Unlike some tau tracers, APN-1607 does not bind off-target to monoamine oxidase. In addition, autopsy studies have confirmed that the PET signal matches tau deposits in PSP, CBD, and Pick’s disease brains (Tagai et al., 2021). Navia said Aprinoia will start a Phase 3 registration study for APN-1607’s use in PSP in 2024.
A TDP-43 Tracer on the Horizon?
In Miami, Navia also reminded the audience of AC Immune’s nascent TDP-43 PET tracer, ACI-19278. This tracer sticks to TDP-43 deposits in FTLD brain sections with 15 nM affinity, with little binding to Aβ and α-synuclein deposits, according to data AC Immune had presented at the 2023 AD/PD conference in Gothenberg, Sweden. Among the different known varieties of TDP-43 fibrils, “Type A” fibrils appear to have had the highest affinity for ACI-19278, Navia said. These occur in FTD brains with GRN mutations, in cases of limbic-predominant age-related TDP-43 encephalopathy (LATE), and as a co-pathology in AD (Aug 2023 news).
At AD/PD, AC Immune had also shown ACI-19278 binding TDP-43 Type B fibrils, which are present in amyotrophic lateral sclerosis and some FTD cases (Dec 2021 news). In nonhuman primates, ACI-19278 entered the brain within minutes and washed out within an hour, making it suitable for live imaging. A clinical trial will start in the second quarter of 2024.
FDG-PET as Differential Diagnostic
With most FTD biomarkers still in early development stages, could an existing technology fill the gap? In Miami, David Jones of the Mayo Clinic in Rochester, Minnesota, made a case for using machine learning to distinguish dementias based on FDG-PET scans. Because the FDG-PET signal falls in brain regions with little metabolic activity, it can flag areas of poor health or degeneration.
The Mayo Clinic developed a web-based application, StateViewer, and trained it on 3,000 PET scans to find patterns representing different diseases. Next, StateViewer was tested on scans from 304 people with AD, 151 with dementia with Lewy bodies, 102 PSP, 97 bvFTD, 77 semantic and 106 logopenic variant primary progressive aphasia, 75 posterior cortical atrophy, 38 semantic dementia, 27 CBD, and 1,706 healthy controls. In this study, the app’s diagnostic accuracy varied from a low of 0.90 for bvFTD to a high of 0.99 for semantic dementia. An earlier version of these data, comprising fewer cases, was presented at the 2022 AAIC in Amsterdam (Barnard et al., 2022).
How does machine learning improve on visual reads? Jones said StateViewer integrates global patterns of change, which are difficult for the human eye to pick out. For example, DLB and PCA look similar to the human eye on FDG scans. Both have the so-called “cingulate island” sign, i.e., a sparing of metabolism in the cingulate relative to dampening in the surrounding precuneus and cuneus. StateViewer easily distinguishes the two diseases based on frontal cortex activity, which is down in the former and up in the latter, Jones said.
Some Holloway attendees were interested in following up on how machine learning can improve diagnosis, and Jones said the software will be widely available soon.
Going Bigger Despite some progress, much more is needed. Billy Dunn, who left the U.S. Food and Drug Administration in February 2023 and is now on the board of Prothena Biosciences, believes the FTD field needs to move toward a biologic definition of disease with a staging system, such as AD has. To do this, promising biomarkers will have to be studied in observational cohorts to find out how they change over the course of the disease, similar to the ADNI and DIAN studies in AD and PPMI in Parkinson’s. In this way, biomarkers will become able to track progression or predict outcomes. Lacking such data, the current batch of markers will be useful as enrollment criteria to select participants who have the pathology being studied, Dunn said. Some such cohorts are in place, e.g. GENFI and ALLFTD, but FTD biomarkers for them are still emerging. Observational FTD cohorts also need more participants and, at the summit, extensive discussion revolved around ways to increase recruitment.
Attendees left the summit charged up. Olivier Piguet at the University of Sydney, paraphrasing a previous line from Leonard Petrucelli at the Mayo Clinic in Jacksonville, Florida, said, “A few years ago we were crawling. Now we’re walking. Soon we may be running.”—Madolyn Bowman Rogers
Tagai K, Ono M, Kubota M, Kitamura S, Takahata K, Seki C, Takado Y, Shinotoh H, Sano Y, Yamamoto Y, Matsuoka K, Takuwa H, Shimojo M, Takahashi M, Kawamura K, Kikuchi T, Okada M, Akiyama H, Suzuki H, Onaya M, Takeda T, Arai K, Arai N, Araki N, Saito Y, Trojanowski JQ, Lee VM, Mishra SK, Yamaguchi Y, Kimura Y, Ichise M, Tomita Y, Zhang MR, Suhara T, Shigeta M, Sahara N, Higuchi M, Shimada H.
High-Contrast In Vivo Imaging of Tau Pathologies in Alzheimer's and Non-Alzheimer's Disease Tauopathies.
Neuron. 2021 Jan 6;109(1):42-58.e8. Epub 2020 Oct 29
PubMed.
Comments
No Available Comments
Make a Comment
To make a comment you must login or register.