In this study, we observed that the CSF YKL-40 changes detected in AD or FTD patients in this and previous studies were not detected in AD and FTLD pathological areas. We observed that YKL-40 levels are remarkably higher in CSF compared to brain. Interestingly, we found increased YKL-40 immunoreactivity in cases with AD-CAA pathology, suggesting a relationship of brain YKL-40 with CAA pathophysiology and/or vascular pathology.
CSF biomarker levels often reflect pathophysiological changes in the brain, as is the case for the classical AD CSF biomarkers: Aβ and hyperphosphorylated tau that reflect brain amyloidosis and tangle formation, respectively [
48,
49]. However, AD is a multifaceted disorder in which multiple processes beyond amyloid and Tau are known to contribute to disease pathogenesis [
50], including immunity [
51]. Additional markers that reflect the neuroinflammatory changes underlying dementia might be useful not only to better define individual patients’ phenotype [
52] but also to monitor treatment responses of anti-inflammatory drugs [
11]. YKL-40 is involved in the immune system response [
16,
33] and human neuropathological studies have shown an astrocytical YKL-40 immunoreactivity in cases with different neurological disorders [
17,
22,
32,
34,
53]. Thus, we hypothesized that the elevated levels of YKL-40 in CSF of AD [
17‐
24] and FTLD [
20,
21,
23,
26‐
29] patients reflect neuroinflammatory changes in the brain areas that are typically affected in these dementia types. In agreement with previous studies, we observed that YKL-40 immunoreactivity was mainly found in clusters of glial cells (likely astrocytes) [
17,
22,
32], and to some extent also in neurons. We observed that YKL-40 immunoreactivity was overall low with many cases (>50%) showing none or few YKL-40 positive cells. This is in line with a previous report showing that only 10% of all GFAP-astrocytes were positive for YKL-40 [
32]. In agreement with a previous study, we also observed a prominent YKL-40 immunoreactivity around the cerebral vessels in AD cases with CAA pathology, suggesting a role of YKL-40 with vascular function [
22]. However, while previous studies reported increased YKL-40 immunoreactivity in AD or FTLD cases [
22,
32], we observed, using complementary methods, that the YKL-40 levels were similar in AD and FTLD compared to non-demented controls. Regarding AD, previous studies focused on frontal cortex areas while here we examined mainly the temporal cortex. The different regions analyzed could partly explain the discrepancies observed, especially considering that YKL-40 expression levels show some regional differences within the brain [
22,
53]. Still, we did not observe any tendency in the small set of samples from AD frontal cortex. Noteworthy, our AD cohort (temporal cortex) is considerably large including more than 50 AD cases and controls, which increases the statistical power of our study. Regarding FTD, previous studies included only FTLD-Tau cases [
32], while here we additionally included FTLD cases with TDP-43 pathology. However, we did not detect any differences when the FTLD pathological subtypes (i.e., FTLD-Tau and FTLD-TDP) were analyzed separately. It is unlikely that the observed immunohistochemically discrepancies are explained by the sample size as the FTLD group was comparable to previous studies. Considering the overall low expression pattern of YKL-40, the semi-quantitative methods applied in the different studies may partly explain the discrepancies across studies. We here analyzed the complete slide and dichotomized the outcomes into either positive or negative, and thus we might have missed subtle changes of YKL-40 immunoreactivity. Quantification of the intensity of YKL-40 in a subset of cases also showed similar results. Furthermore, we employed additional quantitative technologies, including the same immunoassay that is widely used for CSF analysis, which showed again no differences on YKL-40 levels between AD or FTLD and controls post-mortem brain. It is worth noting that a recent study showed no difference in
CHI3L1 mRNA post-mortem tissue from early-onset AD cases, which was attributed to the younger age of these patients [
54]. In line with our findings, a recent mass-spectrometry-based study detected YKL-40 changes in the CSF of AD patients but not in post-mortem tissue [
55]. Interestingly, using paired CSF-brain samples, we observed that CSF YKL-40 levels are 8-times higher than those detected in post-mortem frontal cortex. Strikingly, we observed a non-significant but moderate inverse association of YKL-40 between these two matrices but considering the low sample size of the paired samples (
n = 9), these correlation findings should be interpreted with caution.
Our data thus overall suggest that the YKL-40 changes observed in CSF of dementia patients does not reflect changes of this neuroinflammatory protein in the brain areas that are typically affected. Interestingly, similar results have been observed previously for α-synuclein which was increased in CSF of CJD patients but not in post-mortem brain tissue [
56]. CSF protein levels are dynamic and may change over time depending on the disease stage, as previously observed in longitudinal CSF biomarker studies performed in familial AD patients [
57]. Thus, it could be possible that the YKL-40 changes observed in ante-mortem CSF in various reports might not be observed at the end stage of the disease within brain tissue. However, longitudinal studies have shown that CSF YKL-40 levels increased continuously with disease progression, suggesting that normalization of CSF YKL-40 levels with advancing stage may not explain the lack of differences in tissue [
58,
59]. We neither observed an association of YKL-40 immunoreactivity with advanced pathological stages in brain tissue (i.e., Thal or Braak stages). The absence of YKL-40 changes in AD and FTLD post-mortem tissue together with the prominent reactivity of YKL-40 observed in AD with CAA pathology might indicate that CSF YKL-40 changes could be associated with the peripheral blood compartment. This is, however, less likely since YKL-40 levels in the blood are lower and do not correlate with those in CSF and its levels remain unchanged in the blood of dementia patients [
27,
60]. YKL-40 is a protein secreted by various cell types including astrocytes [
33,
61]. Thus, YKL-40 may quickly diffuse from cells to the extracellular space and the CSF, which may ultimately hamper its detection in post-mortem brain. The origin of the increased CSF YKL-40 levels consistently observed in AD and FTD patients is thus still not clear. CSF YKL-40 changes may originate from an alternative area not analyzed in the current study, such as the choroid plexus, which is involved in the production and regulation of CSF and has shown to express YKL-40 at least during brain development [
62]. The choroid plexus could play an important role in facilitating the inflammatory process in the central nervous system. This is supported by others that show the presence of immune cells [
63,
64] and an up-regulation of pro-inflammatory cytokines and chemokines in the choroid plexus of AD patients [
65], suggesting an involvement of this area within the neuroinflammatory process associated with aging and AD [
66,
67].
This study is not without limitations. The sample size of AD-CAA cases, as well as those with paired CSF-brain tissue, was small, and thus, such data should be confirmed in larger cohorts. In addition, our paired CSF-brain samples did not include AD patients or non-demented controls, and thus we can not exclude that the correlation of YKL-40 in brain and CSF might be different in these groups. However, the CSF YKL-40 values of the paired samples covered a wide concentration range (200–700 ng/mL) including YKL-40 values comparable to those previously detected in the control or AD groups. Furthermore, our study is limited to the frontal and temporal cortex, and thus other brain areas (e.g., choroid plexus) may contribute to the strong YKL-40 changes observed in CSF. We also acknowledge that we could not compare YKL-40 across different diseases due to the difference in brain regions. The strengths of our study are the use of complementary methods ensuring reliable measurements of YKL-40 protein and the high number of cases analyzed.