ATN classifications in a mixed cohort of frontotemporal lobar degeneration (FTLD) and Alzheimer’s disease (AD) pathology using cerebrospinal fluid neurofilament light chain (NFL)

Abstract

AbstractBackgroundThe ATN framework combines cerebrospinal fluid (CSF) amyloidβ 1‐42 (Aβ1‐42), phosphorylated tau (p‐tau), and total tau (t‐tau) to diagnose individuals along the continuum of Alzheimer’s disease (AD). While ATN profiles indicative of AD are highly accurate, profiles indicating normal, non‐AD pathology, or concurrent AD and non‐AD pathology are inconsistent, due in part to the insensitivity of t‐tau to non‐AD pathologies, such as frontotemporal lobar degeneration (FTLD). We test a modified framework that replaces t‐tau with neurofilament light chain (NFL), a CSF marker of axonal degeneration.MethodWe compare ATN classifications using t‐tau (ATNt) and NFL (ATNf) in an pathology‐confirmed autopsy cohort of 72 AD (3 with concurrent FTLD) and 26 FTLD (14 with concurrent low‐high AD). In an independent living sample, receiver operating characteristic (ROC) analyses defined an optimal NFL cut‐point using 18 hereditary FTLD (10 C9ORF72, 3 MAPT, 2 progranulin, and 3 TARDBP) and 27 matched control subjects without cognitive impairment. Published cut‐points (Shaw et al., 2009) defined A (Aβ1‐42<192 pg/mL), T (p‐tau>23 pg/mL), and Nt (t‐tau>93 pg/mL) status, and the above ROC‐threshold (NFL>1003 pg/mL) defined Nf status to create a CSF‐based ATNt and ATNf profile for each patient.ResultATNt and ATNf were equally likely to identify Alzheimer’s continuum disease based on Aβ1‐42. Fisher’s test indicated no difference in Nf‐positivity (41 of 72) and Nt‐positivity (44 of 72) for AD pathology (p=0.73). ATNf correctly identified 25 of 26 FTLD cases as non‐AD pathology; 21 FTLD cases were classified as non‐AD (A‐T‐Nf+/A‐T+Nf+), 4 as non‐AD and concurrent AD (A+T‐Nf+), and 1 misclassified as normal (A‐T‐Nf‐). Fisher’s test showed that ATNf was significantly more likely to identify FTLD as non‐AD pathology than ATNt (OR = 152.9, CI = 16 – 7639, p < 0.001); under ATNt, only 3 FTLD cases were classified as non‐AD (A‐T‐Nt+/A‐T+Nt+), while 19 were misclassified as normal (A‐T‐Nt‐) and 4 as early AD pathologic change (A+T‐Nt‐).ConclusionATNf designations using NFL were more accurate than t‐tau when assessing a mixed AD‐FTLD pathology cohort. Future biomarker strategies to detect both AD and non‐AD neuropathologies, such as FTLD, should consider alternative markers for N.