The heterogeneity of neurodegenerative diseases is a key confound to disease understanding and treatment development, as study cohorts typically include multiple phenotypes on distinct disease trajectories. Here we introduce a machine-learning technique—Subtype and Stage Inference (SuStaIn)—able to uncover data-driven disease phenotypes with distinct temporal progression patterns, from widely available cross-sectional patient studies. Results from imaging studies in two neurodegenerative diseases reveal subgroups and their distinct trajectories of regional neurodegeneration. In genetic frontotemporal dementia, SuStaIn identifies genotypes from imaging alone, validating its ability to identify subtypes; further the technique reveals within-genotype heterogeneity. In Alzheimer’s disease, SuStaIn uncovers three subtypes, uniquely characterising their temporal complexity. SuStaIn provides fine-grained patient stratification, which substantially enhances the ability to predict conversion between diagnostic categories over standard models that ignore subtype (p = 7.18 × 10⁻⁴) or temporal stage (p = 3.96 × 10⁻⁵). SuStaIn offers new promise for enabling disease subtype discovery and precision medicine.

Frontotemporal dementia is a highly heritable neurodegenerative disorder. In about a third of patients, the disease is caused by autosomal dominant genetic mutations usually in one of three genes: progranulin (GRN), microtubule-associated protein tau (MAPT), or chromosome 9 open reading frame 72 (C9orf72). Findings from studies of other genetic dementias have shown neuroimaging and cognitive changes before symptoms onset, and we aimed to identify whether such changes could be shown in frontotemporal dementia. We recruited participants to this multicentre study who either were known carriers of a pathogenic mutation in GRN, MAPT, or C9orf72, or were at risk of carrying a mutation because a first-degree relative was a known symptomatic carrier. We calculated time to expected onset as the difference between age at assessment and mean age at onset within the family. Participants underwent a standardised clinical assessment and neuropsychological battery. We did MRI and generated cortical and subcortical volumes using a parcellation of the volumetric T1-weighted scan. We used linear mixed-effects models to examine whether the association of neuropsychology and imaging measures with time to expected onset of symptoms differed between mutation carriers and non-carriers. Between Jan 30, 2012, and Sept 15, 2013, we recruited participants from 11 research sites in the UK, Italy, the Netherlands, Sweden, and Canada. We analysed data from 220 participants: 118 mutation carriers (40 symptomatic and 78 asymptomatic) and 102 non-carriers. For neuropsychology measures, we noted the earliest significant differences between mutation carriers and non-carriers 5 years before expected onset, when differences were significant for all measures except for tests of immediate recall and verbal fluency. We noted the largest Z score differences between carriers and non-carriers 5 years before expected onset in tests of naming (Boston Naming Test -0·7; SE 0·3) and executive function (Trail Making Test Part B, Digit Span backwards, and Digit Symbol Task, all -0·5, SE 0·2). For imaging measures, we noted differences earliest for the insula (at 10 years before expected symptom onset, mean volume as a percentage of total intracranial volume was 0·80% in mutation carriers and 0·84% in non-carriers; difference -0·04, SE 0·02) followed by the temporal lobe (at 10 years before expected symptom onset, mean volume as a percentage of total intracranial volume 8·1% in mutation carriers and 8·3% in non-carriers; difference -0·2, SE 0·1). Structural imaging and cognitive changes can be identified 5-10 years before expected onset of symptoms in asymptomatic adults at risk of genetic frontotemporal dementia. These findings could help to define biomarkers that can stage presymptomatic disease and track disease progression, which will be important for future therapeutic trials. Centres of Excellence in Neurodegeneration.

The factors driving clinical heterogeneity in Alzheimer’s disease are not well understood. This study assessed the relationship between amyloid deposition, glucose metabolism and clinical phenotype in Alzheimer’s disease, and investigated how these relate to the involvement of functional networks. The study included 17 patients with early-onset Alzheimer’s disease (age at onset <65 years), 12 patients with logopenic variant primary progressive aphasia and 13 patients with posterior cortical atrophy [whole Alzheimer’s disease group: age = 61.5 years (standard deviation 6.5 years), 55% male]. Thirty healthy control subjects [age = 70.8 (3.3) years, 47% male] were also included. Subjects underwent positron emission tomography with ¹¹C-labelled Pittsburgh compound B and ¹⁸F-labelled fluorodeoxyglucose. All patients met National Institute on Ageing–Alzheimer’s Association criteria for probable Alzheimer’s disease and showed evidence of amyloid deposition on ¹¹C-labelled Pittsburgh compound B positron emission tomography. We hypothesized that hypometabolism patterns would differ across variants, reflecting involvement of specific functional networks, whereas amyloid patterns would be diffuse and similar across variants. We tested these hypotheses using three complimentary approaches: (i) mass-univariate voxel-wise group comparison of ¹⁸F-labelled fluorodeoxyglucose and ¹¹C-labelled Pittsburgh compound B; (ii) generation of covariance maps across all subjects with Alzheimer’s disease from seed regions of interest specifically atrophied in each variant, and comparison of these maps to functional network templates; and (iii) extraction of ¹¹C-labelled Pittsburgh compound B and ¹⁸F-labelled fluorodeoxyglucose values from functional network templates. Alzheimer’s disease clinical groups showed syndrome-specific ¹⁸F-labelled fluorodeoxyglucose patterns, with greater parieto-occipital involvement in posterior cortical atrophy, and asymmetric involvement of left temporoparietal regions in logopenic variant primary progressive aphasia. In contrast, all Alzheimer’s disease variants showed diffuse patterns of ¹¹C-labelled Pittsburgh compound B binding, with posterior cortical atrophy additionally showing elevated uptake in occipital cortex compared with early-onset Alzheimer’s disease. The seed region of interest covariance analysis revealed distinct ¹⁸F-labelled fluorodeoxyglucose correlation patterns that greatly overlapped with the right executive-control network for the early-onset Alzheimer’s disease region of interest, the left language network for the logopenic variant primary progressive aphasia region of interest, and the higher visual network for the posterior cortical atrophy region of interest. In contrast, ¹¹C-labelled Pittsburgh compound B covariance maps for each region of interest were diffuse. Finally, ¹⁸F-labelled fluorodeoxyglucose was similarly reduced in all Alzheimer’s disease variants in the dorsal and left ventral default mode network, whereas significant differences were found in the right ventral default mode, right executive-control (both lower in early-onset Alzheimer’s disease and posterior cortical atrophy than logopenic variant primary progressive aphasia) and higher-order visual network (lower in posterior cortical atrophy than in early-onset Alzheimer’s disease and logopenic variant primary progressive aphasia), with a trend towards lower ¹⁸F-labelled fluorodeoxyglucose also found in the left language network in logopenic variant primary progressive aphasia. There were no differences in ¹¹C-labelled Pittsburgh compound B binding between syndromes in any of the networks. Our data suggest that Alzheimer’s disease syndromes are associated with degeneration of specific functional networks, and that fibrillar amyloid-β deposition explains at most a small amount of the clinico-anatomic heterogeneity in Alzheimer’s disease.

Objective: To evaluate cerebrospinal fluid ( CSF ) and serum neurofilament light chain (NfL) levels in genetic frontotemporal dementia ( FTD ) as a potential biomarker in the presymptomatic stage and during the conversion into the symptomatic stage. Additionally, to correlate NfL levels to clinical and neuroimaging parameters. Methods: In this multicenter case–control study, we investigated CSF NfL in 174 subjects (48 controls, 40 presymptomatic carriers and 86 patients with microtubule‐associated protein tau ( MAPT ), progranulin ( GRN ), and chromosome 9 open reading frame 72 ( C9orf72 ) mutations), and serum NfL in 118 subjects (39 controls, 44 presymptomatic carriers, 35 patients). In 55 subjects both CSF and serum was determined. In two subjects CSF was available before and after symptom onset (converters). Additionally, NfL levels were correlated with clinical parameters, survival, and regional brain atrophy. Results: CSF NfL levels in patients (median 6762 pg/mL, interquartile range 3186–9309 pg/mL) were strongly elevated compared with presymptomatic carriers (804 pg/mL, 627–1173 pg/mL, P < 0.001), resulting in a good diagnostic performance to discriminate both groups. Serum NfL correlated highly with CSF NfL ( r _s = 0.87, P < 0.001) and was similarly elevated in patients. Longitudinal samples in the converters showed a three‐ to fourfold increase in CSF NfL after disease onset. Additionally, NfL levels in patients correlated with disease severity, brain atrophy, annualized brain atrophy rate and survival. Interpretation: NfL in both serum and CSF has the potential to serve as a biomarker for clinical disease onset and has a prognostic value in genetic FTD .