Study Clarifies Associations Between Alzheimer Biomarkers and MRI Measures

April 8, 2021
Victoria Johnson
Victoria Johnson

Victoria Johnson, Assistant Editor for NeurologyLive, joined the MJH Life Sciences team in October 2020. Follow her on Twitter @VictoriaJNeuro or email her at

Researchers studied associations between CSF neurofilament light, neurogranin, and t-tau with measures of cortical thickness, white matter hyperintensities, and microstructural integrity.

A recent study evaluated Alzheimer disease (AD) biomarkers such as cerebrospinal (CSF) neurofilament light (NfL), neurogranin (Ng), and total-tau (t-tau) to clarify how these markers are differentially mapped to magnetic resonance imaging (MRI) measures of cortical thickness, microstructural integrity (corpus callosum and cingulum fractional anisotropy [FA]), and white matter hyperintensities (WMH).

The research group found that higher CSF NfL was associated with decreasing microstructural integrity and WMH. Higher t-tau was associated with decreasing temporal lobe and AD meta region of interest (ROI) cortical thickness.

“It is important to understand what information different neurodegenerative biomarkers provide to aid in both diagnosis and prognosis, especially because each marker is likely to be reflective of a different composite of pathological processes,” wrote first author Michelle M. Mielke, PhD, professor of epidemiology and neurology, and researcher, Mayo Clinic, and colleagues.

Mielke and colleagues analyzed data from 536 participants without dementia from the Mayo Clinic Study of Aging. CSF NfL, Ng, t-tau, amyloid beta (Aβ) 42 and longitudinal MRI scans were performed. They used linear mixed models to assess longitudinal associations between CSF markers and MRI changes.

READ MORE: Retinal Biomarkers May Serve as Alzheimer Disease Detection Tool

Participants’ mean age was 74.9 years (standard deviation [SD], 7.3) and 304 (57%) were men. Out of all participants, 154 (29%) had an apolipoprotein E (APOE) ɛ4 allele and 70 (13%) had a clinical diagnosis of mild cognitive impairment (MCI). The participants had an average follow-up of 2.4 years (SD, 1.8) with 2.2 (SD, 1.0) visits.

Mielke and colleagues found that CSF Ng was strongly associated with t-tau (Spearman’s rho = 0.831; P <.0001) and t-tau (Spearman’s rho = 0.477; P <.0001). CSF NfL was moderately associated with Ng (Spearman’s rho = 0.321; P <.0001) and t-tau (Spearman’s rho = 0.477; P < .0001).

Multivariable analyses revealed that that higher CSF NfL levels were cross-sectionally associated with lower AD meta- region of interest (ROI) cortical thickness (b, –0.149 [standard error (SE), 0.044]; P <.001) and with lower cortical thickness in the frontal (b, –0.224 [SE, 0.046]; P <.001), parietal (b, –0.163 [SE, 0.043]; P <.001), and temporal (b, –0.134 [SE, 0.042]; P <.002) lobes.

There were few cross-sectional associations between either CSF t-tau or Ng with most neuroimaging measures, with the exception that higher levels of t-tau (b, 0.209 [SE, 0.076]; P <.006) and Ng (b, 0.216 [SE, 0.070]; P <.002) were associated with higher corpus callosum FA. Higher CSF t-tau levels were also associated with lower temporal lobe thickness in amyloid positive participants compared to amyloid negative in multivariable analyses (b, –0.215 [SE, 0.081]; P = .002). 

Associations between CSF t-tau or Ng and neuroimaging markers did not differ in those with or without MCI after adjusting for age, sex, and CSF Aβ42. However, higher levels of CSF NfL were associated with reduced cortical thickness in participants with MCI in the frontal (b, –0.524 [SE, 0.131]; P <.001), parietal (b, –0.437 [SE, 0.125]; P <.001), and temporal lobes (b, -0.482 [SE, 0.121]; P <.001) and in an AD meta-ROI (b, –0.653 [SE, 0.125]; P <.001). 

Higher CSF t-tau was longitudinally associated with declines in temporal lobe cortical thickness (b, –0.189 [SE, 0.068]; P = .006) and the AD signature region (b, –0.281 [SE, 0.073]; P <.001). Higher CSF NfL was associated with declines in the corpus callosum (b, – 0.283 [SE, 0.096]; P = .003) and cingulum (b, –0.308 [SE, 0.096]; P = .006) FA and WMH (b, 0.149 [SE, 0.055]; P = .007). No associations were seen between CSF Ng and longitudinal neuroimaging measures. Restricted analyses only including 268 participants with all neuroimaging modalities confirmed associations in all above analyses.

Mielke and colleagues found that associations between CSF t-tau or NfL and longitudinal neuroimaging markers did not differ by CSF Aβ42 status after adjusting for age and sex. Higher CSF Ng was associated with greater declines in temporal lobe thickness (b, –0.464 [SE, 0.140]; P <.001), AD meta-ROI thickness (b, –0.471 [SE, 0.151]; P = .002), and cingulum FA (b, –0.662 [SE, 0.240]; P = .006) in amyloid-positive individuals compared to amyloid-negative individuals. Restricting the analyses to participants with all neuroimaging modalities yielded similar results, with the exception that the interaction between CSF Ng and CSF Aβ42 for cingulum FA was no longer significant (P = .015). 

Lastly, increasing levels of CSF t-tau (b, –0.579 [SE, 0.218]; P = .008) and Ng (b, –0.664 [SE, 0.231]; P = .007) were associated with greater reduction in AD meta-ROI cortical thickness longitudinally in participants with MCI compared to those without. No differences were seen between CSF NfL and all neuroimaging outcomes between participants with or without diagnosed MCI.

“The present study contributes further insight regarding information each of these three candidate CSF markers may provide and how they can best be used for clinical and research purposes. Of the three, CSF Ng appears to be the best marker of AD-related pathological neurodegeneration (NfL, Ng, and t-tau) and will be most helpful for AD diagnosis and AD-related prognosis,” Mielke and colleagues wrote.

“Although CSF t-tau provides some indication of AD pathology, it appears to be less specific than Ng. Last, CSF NfL is non-specific to AD pathology. The measurement of CSF NfL has the added benefit of providing information regarding non-AD related pathologies, potentially due to vascular and other causes, and will add information about overall prognosis,” they concluded.

Mielke MM, Przybelski SA, Lesnick TG, et al. Comparison of CSF neurofilament light chain, neurogranin, and tau to MRI markers. Alzheimers Dement. Published online March 4, 2021. doi: 10.1002/alz.12239