Bahl, Riley

AbstractBackgroundClinical trials are underway to treat familial frontotemporal lobar degeneration (f‐FTLD). This is a rare disease, and a limited number of mutation carriers have been identified; thus, efficient trial design is critical. Multimodal, latent disease progression models (DPM) can estimate time to symptom onset and define the temporal ordering of biomarker changes. DPMs can also be leveraged to select endpoints and potentially supplement analyses by integrating historical data. Recent draft FDA guidance for gene therapy trials in neurological disease supports these novel approaches to clinical trials.MethodParticipants included 1,049 members of families affected by f‐FTLD, due to mutations in GRN, MAPT, or C9orf72 genes, who were enrolled in ALLFTD or GENFI. A Bayesian repeated measures model incorporated multimodal data to estimate disease progression, conditional on latent disease age (proximity to symptom onset), in 677 mutations carriers (GRN (n=233), MAPT (n=151) and C9orf72 (n=293)). Family members without pathogenic mutations were used as the reference group. Mean follow‐up was 1.1 (SD=1.1) years. Jointly modeled longitudinal variables included neuropsychological scores, CDR®+NACC‐FTLD Box Score, MRI volumes of brain regions affected by f‐FTLD, and plasma levels of neurofilament light chain (NfL).ResultDisease progression curves were similar across ALLFTD and GENFI cohorts. Plasma NfL elevations occurred earliest, up to 10 years before symptom onset, and NfL was the most powerful endpoint in the asymptomatic stage. MRI abnormalities occurred next, closer to symptom onset. The earliest MRI changes relative to symptom onset were observed in C9orf72+. GRN mutation carriers showed the most rapid acceleration in all biomarkers, and this acceleration occurred in close proximity to symptom onset. Neuropsychological measures and CDR®+NACC‐FTLD Box Score were among the most promising endpoints in the symptomatic stage. Trial simulations indicated that using latent disease age as an enrollment criterion would allow some asymptomatic mutation carriers to be enrolled without sacrificing power.ConclusionSimilarity in disease progression across ALLFTD and GENFI participants suggests these models will apply to international trials. Model‐derived estimates of disease progression curves indicate that endpoint selection should be specific to disease stage and mutation, and DPMs would facilitate greater participant enrollment.

AbstractBackgroundIn preparation for treatment trials in frontotemporal dementia (FTD), sensitive assessments to detect early clinical deficits and track longitudinal changes are needed. Remote digital data collection using smartphones may improve sensitivity to detect and follow abnormalities by allowing novel methods of data capture and more frequent assessments. Furthermore, remote evaluations can reduce patient and staff burden. Despite these benefits, little is known about the reliability and validity of self‐administered cognitive testing completed on smartphones.MethodWe investigated self‐administered smartphone cognitive tests from the ALLFTD Mobile App in a mixed sample of 45 participants, including 10 functionally intact older adults, 24 participants from families harboring pathogenic FTD mutations (asymptomatic n=14; mildly symptomatic n=10;), and 11 symptomatic FTD‐spectrum participants. A subset (n=15) was missing CDR®+NACC‐FTLD data but consensus diagnoses were available. Participants were asked to complete five tests that were repeated three times over eight days. Tests included versions of N‐back, Stroop, Flanker, and Go/No‐Go paradigms, and an adaptive spatial working memory test. Reliability was assessed using split‐half reliability (first testing session, 100 simulations) and intraclass correlations (ICC) across the three testing sessions. Linear models were fitted to evaluate the association of test performance with age and disease severity (CDR®+NACC‐FTLD Global and Box Scores), covarying for sex and education level. Models evaluating age associations excluded symptomatic participants.ResultParticipants completed 72.4% of available instruments. Split‐half reliability (range: 0.76 to 0.97) and ICCs (0.70 to 0.92) suggested adequate‐to‐excellent internal consistency and test‐retest reliability. Across the five tests, negative associations were observed with CDR®+NACC‐FTLD Box Scores (ßs = ‐0.24 to ‐0.77), and in clinically asymptomatic individuals, negative associations were observed with age (ßs = ‐0.43 to ‐0.88). Performance on all measures was worse in symptomatic participants compared to asymptomatic (p‐values<.05). Mildly symptomatic cases (CDR®+NACC‐FTLD=0.5) performed worse on Stroop, Flanker, and spatial working memory than asymptomatic participants (p‐values<.03). Flanker and Stroop tasks consistently showed the strongest reliability and validity findings.ConclusionSmartphone adaptations of common neuropsychological tests produced promising preliminary results. Strong associations with disease severity and age suggest these metrics may be valid for detecting clinically relevant cognitive changes, even in mildly symptomatic FTD.

AbstractBackgroundClinicopathologic studies reported that the NACC Uniform Dataset (UDS) 2.0 core measures do not adequately differentiate patients who ultimately died of Frontotemporal Lobar Degeneration (FTLD) from Alzheimer’s disease (AD). The FTLD Module (FTLD‐MOD) assesses core features of behavioral variant FTD (bvFTD) and primary progressive aphasia (PPA). Initial investigations suggest that the FTLD‐MOD improves discrimination between biomarker‐supported bvFTD and typical late‐onset AD, and that clinically diagnosed PPA participants performed worse than bvFTD on all FTLD‐MOD language tasks. However, the utility of these measures in the differential diagnosis of AD vs. major FTLD syndromes in young‐onset cases remains underexplored.MethodWe examined participants with age‐of‐onset < 65 years in the ARTFL‐LEFFTDS Longitudinal Frontotemporal Lobar Degeneration (ALLFTD) consortium and the Longitudinal Early Onset AD Study (LEADS). 115 bvFTD, 53 semantic variant PPA (svPPA), 29 non‐fluent variant PPA (nfvPPA) and 124 early‐onset AD (eoAD) cases completed the UDS3 and FLTD‐MOD. Linear regression analyses measured group differences in FLTD‐MOD raw scores and UDS3 neuropsychological z‐scores after accounting for group differences in demographic variables and global severity ratings. Alpha was 0.05 with Bonferroni corrections.ResultsParticipants with eoAD demonstrated the most impaired scores on tasks of attention, working memory, and processing speed, including Digit Span Backward (z=‐1.87) and Trails A (z=‐3.82). They also demonstrated difficulty on FTLD‐MOD language tasks with a working memory requirement (sentence repetition raw=3.68/5) and, as expected, memory (story delayed recall z=‐2.4z). svPPA performed the worst on FTLD‐MOD and UDS3 tasks assessing semantic memory, especially naming (all p’s ≤ 0.03). nfvPPA performed worst on sentence reading, sentence repetition, and letter fluency (all p’s ≤ 0.02). Consistent with previous work, these measures differentiate between bvFTD and PPA variants. However, relative to eoAD, bvFTD was equivalent (Trails B: bvFTDz=‐3.8, eoADz=‐3.82; p=0.12) or outperformed (sentence repetition: bvFTD=4.22/5, eoAD=3.68/5; p<0.001) the eoAD group on all tasks.ConclusionsWe undertook this analysis to compare performance of the major young‐onset sporadic AD and FTLD clinical syndromes in the ALLFTD and LEADS studies. Results demonstrate relatively distinct cognitive profiles among eoAD, svPPA, and nfvPPA, but not in bvFTD. Future analyses will include FTLD‐MOD behavioral symptom questionnaires to assess their utility in differentiating bvFTD.