Seed amplification and neurodegeneration marker trajectories in individuals at risk of prion disease

Abstract

Human prion diseases are remarkable for long incubation times followed typically by rapid clinical decline. Seed amplification assays and neurodegeneration biofluid biomarkers are remarkably useful in the clinical phase, but their potential to predict clinical onset in healthy people remains unclear. This is relevant not only to the design of preventive strategies in those at-risk of prion diseases, but more broadly, because prion-like mechanisms are thought to underpin many neurodegenerative disorders. Here, we report the accrual of a longitudinal biofluid resource in patients, controls and healthy people at risk of prion diseases, to which ultrasensitive techniques such as real-time quaking-induced conversion (RT-QuIC) and single molecule array (Simoa) digital immunoassays were applied for preclinical biomarker discovery. We studied 648 CSF and plasma samples, including 16 people who had samples taken when healthy but later developed inherited prion disease (IPD) ('converters'; range from 9.9 prior to, and 7.4 years after onset). Symptomatic IPD CSF samples were screened by RT-QuIC assay variations, before testing the entire collection of at-risk samples using the most sensitive assay. Glial fibrillary acidic protein (GFAP), neurofilament light (NfL), tau and UCH-L1 levels were measured in plasma and CSF. Second generation (IQ-CSF) RT-QuIC proved 100% sensitive and specific for sporadic Creutzfeldt-Jakob disease (CJD), iatrogenic and familial CJD phenotypes, and subsequently detected seeding activity in four presymptomatic CSF samples from three E200K carriers; one converted in under 2 months while two remain asymptomatic after at least 3 years' follow-up. A bespoke HuPrP P102L RT-QuIC showed partial sensitivity for P102L disease. No compatible RT-QuIC assay was discovered for classical 6-OPRI, A117V and D178N, and these at-risk samples tested negative with bank vole RT-QuIC. Plasma GFAP and NfL, and CSF NfL levels emerged as proximity markers of neurodegeneration in the typically slow IPDs (e.g. P102L), with significant differences in mean values segregating healthy control from IPD carriers (within 2 years to onset) and symptomatic IPD cohorts; plasma GFAP appears to change before NfL, and before clinical conversion. In conclusion, we show distinct biomarker trajectories in fast and slow IPDs. Specifically, we identify several years of presymptomatic seeding positivity in E200K, a new proximity marker (plasma GFAP) and sequential neurodegenerative marker evolution (plasma GFAP followed by NfL) in slow IPDs. We suggest a new preclinical staging system featuring clinical, seeding and neurodegeneration aspects, for validation with larger prion at-risk cohorts, and with potential application to other neurodegenerative proteopathies.

Keywords: GFAP; NfL; RT-QuIC; inherited; prion.

Figure 1

IPD-AR, iCJD-AR and IPD converter biofluid sample archive. This graph plots all the samples (plasma only, CSF only or matched plasma and CSF) analysed in this study, grouped by age category (<40, 40–49, 50–59 and >60) on the x-axis to obscure identities, with each minor tick mark after the start of each age category reflecting an interval of 1 year. A total of 12 years are covered per age group as the longest follow-up is over 11 years, in order to avoid overlapped timelines. The first (or only) sample from each individual is collapsed to the start of each age category to preserve anonymity. Samples from the same individual are joined by a horizontal line if more than one sample was collected; thick black horizontal lines denote onset of clinical conversion. Converters are grouped together in the upper shaded part of the graph. For converters where only one presymptomatic sample exists without any follow-up samples, the subsequent data-point (unfilled inverted triangle marker) joined by line indicates time of death. IPD mutations with fewer than five at-risk individuals were grouped as ‘Other’ to avoid self-identification.

Figure 2

Graphs of select IPD-AR and control samples with positive RT-QuIC results. (A) This is the sole IQ-CSF RT-QuIC positive E200K-AR sample, which recorded fewer than 4/4 wells positive, drawn at 3.37 years from the present time. (B) This is the sole HuPrP P102L RT-QuIC positive sample in the P102L-AR set; this sample was negative when tested with BV RT-QuIC. (C) This non-prion disease (neurodegenerative) CSF sample tested positive with Hu P102L RT-QuIC, but tested negative with IQ-CSF RT-QuIC and BV RT-QuIC. The dotted vertical lines indicate the time cut-offs for the individual assays i.e. 24 h for IQ-CSF RT-QuIC and 50 h for Hu P102L RT-QuIC.

Figure 3

RT-QuIC CSF end point dilutions for E200K-AR and E200K converter samples to calculate SD₅₀/ µl. Each panel series show dilutions of seeding E200K CSF volume by a third; the vertical dotted line indicates the time cut-off, which was 24 h for IQ-CSF RT-QuIC. (A and B) From a single individual drawn at 3.75 and 1.70 years from the present, respectively. (C and D) From a single converter individual 0.6 years apart at 0.2 years to, and 0.4 years after conversion, respectively. The dotted vertical lines indicate the time cut-offs for the individual assays i.e. 24 h.

Figure 4

Simoa N4PB measurements in prion disease and at-risk cohorts. (A) Plasma N4PB levels, where only GFAP and NfL showed statistically significant different mean values between IPD at-risk and disease groups. (B) CSF N4PB levels, where only NfL showed statistically significant different mean values between IPD at-risk and disease groups.

Figure 5

Converter trajectories for plasma GFAP and NfL and CSF NfL. Plasma GFAP (A) and NfL (B) trajectories are grouped into P102L (slow IPD) and E200K + D178N-FFI (fast IPDs) and OPRIs (slow IPDs; includes 5- and 6-OPRI). CSF converter data-points for NfL are shown in C. The horizontal dotted line indicates the 90th percentile of the respective biomarker value in the normal control cohort.

Figure 6

Proposed pre-conversion IPD patterns of biomarker change for fast and slow IPDs. Each stage features expected intensities in PrP-amyloid seeding activity, neurodegeneration markers and clinical aspects, along with ancillary investigations known to herald the onset of conversion (neuropsychometry, and neurophysiology in P102L). Naturally, the small numbers in this study precludes the provision of precise quantitative scales at the present time. (A) Slow IPDs are likely to have an extended window for neurodegenerative markers, making it easier to capture and follow at 6–12 monthly sampling intervals; however, we only have partially sensitive RT-QuIC seeding assays for slow IPDs. (B) Fast IPDs are likely to have a very short and explosive neurodegeneration window, which means it might not be easy to capture and follow at similar sampling intervals; this may be offset by the existence of highly sensitive RT-QuIC assays (E200K only) that may become positive several years before clinical onset. The changes in CSF PrP amyloid seeding are hypothetical, current evidence is limited to a very small number of individuals and samples.