95 Markers of neurodegeneration in subjective cognitive decline the surrounding CSF spaces, which means it reflects hippocampal atrophy as well as global and subcortical atrophy (42). Furthermore, being cognitively normal, most individuals in our sample had an MTA score of 0, which reflects that the variability for this measure is probably too small to be a meaningful N biomarker in such a very early sample. In addition, the correlation coefficients between N biomarkers were in a similar range as the correlation coefficients between N biomarkers on the one hand, and A and T biomarkers on the other hand. This is in line with another study which found moderate correlations between biomarkers of different pathophysiological categories (6). This implies that the underlying neurodegeneration processes are almost as different to each other, as they are different to processes underlying the A and T biomarker category. Overall, the low correlation coefficients illustrate that N biomarkers cannot be used interchangeably in the AT(N) classification. We found that HV, NfL and GFAP predicted clinical progression, and HV predicted MMSE slope, beyond Aβ and p-tau. Former studies that investigated the AT(N) classification often used only one biomarker for A, T and N respectively, and showed that overall, the AT(N) classification was associated with clinical progression and cognitive decline (20, 43-47). From these studies, the predictive value per individual biomarker cannot be discerned and thus cannot be used to choose the optimal N biomarker. Literature regarding the comparison between different N biomarkers is more scarce. There is, however, some support that HV is associated with cognitive decline and progression more strongly than t-tau (13-15). Although in our study, we found t-tau as individual biomarker also predicted clinical progression and cognitive decline, the high correlation with p-tau hampers the addition of t-tau to a model with Aβ and p-tau, making it a less desirable biomarker to use in the AT(N) classification. NfL and GFAP have both been shown to be related to baseline cognition, cognitive decline and clinical progression as individual predictors, but have not yet been studied extensively in comparison to other N biomarkers (3, 48-50). In a former study, we found GFAP was more strongly related to clinical progression and cognitive decline than NfL, which is in line with our current study (4). We found both GFAP and NfL predicted clinical progression beyond Aβ and p-tau, but NfL was not associated with MMSE decline. A potential explanation for this difference in association is that NfL is a better marker for monitoring disease progression while its value does not lie in predicting future cognitive decline (4). Differences could also be related to the fact that clinical progression to MCI or dementia is a binary outcome measure, while MMSE decline is a continuous measure with possibly a higher degree of measurement variation. Clinical progression might be a more sensitive measure with more clinical relevance. In contrast to NfL, GFAP was associated with MMSE decline, although associations were attenuated when additionally adjusting for Aβ and/or p-tau. Of 4
RkJQdWJsaXNoZXIy MjY0ODMw