In this study we evaluated the association of the missense AD risk variants ABI3_rs616338-T and PLCG2_rs72824905-G with both disease risk and neuropathology in autopsy-confirmed cohorts comprised of 973 LBD-NP and 1040 PSP patients. To our knowledge, this is one of the largest studies to date of these variants in these purely neuropathologically diagnosed neurodegenerative diseases. Our findings suggest that the protective role of PLCG2_rs72824905-G may be through a suppressive effect on tau pathology. We also find evidence of ABI3_rs616338-T association with LBD-NP risk in a sub-category of patients with diffuse neocortical or limbic LB, concurrently with moderate or high AD neuropathology, respectively. Finally, in our sizable autopsy-confirmed PSP cohort, we find no evidence of association with disease risk for either variant.
In our previous study , we analyzed both variants in 306 DLB patients of whom 22% were autopsy-confirmed (LBD-NP) and 231 neuropathologically diagnosed PSP patients, in addition to AD, PD and MSA cohorts. The current study achieved a > 3 and > 4.5 fold increase in sample size for LBD-NP and PSP, respectively, compared to our prior work .
Our lack of statistically significant disease risk association with PLCG2_rs72824905-G and our autopsy-confirmed PSP cohort is consistent with the findings from the multi-center van der Lee et al.  study, which analyzed 882 PSP patients of whom 29% were autopsy-confirmed. In a GWAS of 2165 PSP cases (around 58% autopsy-confirmed), no associations were identified at the PLCG2 or ABI3 loci , similar to our results in this study.
Likewise, we did not find statistically significant associations with either variant and disease risk in our combined autopsy-confirmed cohort of 973 LBD-NP patients. This is in contrast to the multi-center van der Lee et al.  study, which detected association between PLCG2_rs72824905-G and reduced disease risk in 1446 DLB patients, of whom 11% had neuropathologic diagnosis. In a GWAS of 1743 DLB patients, of whom 1324 were autopsy-confirmed , no associations were identified at the PLCG2 or ABI3 loci, although significant associations were reported for APOE, SNCA and GBA. More recently, whole exome sequencing of 1118 autopsy-confirmed DLB patients  did not identify significant association with PLCG2_rs72824905-G.
These collective results suggest either that variants in ABI3 and PLCG2 do not play as significant a role in PSP and LBD-NP as in AD or that the existing studies are underpowered. Indeed, the initial report of AD risk associations with these rare variants was based on 37,022 AD cases and 48,402 controls , which is at least an order of magnitude greater than all studies combined for PSP or LBD-NP. Future efforts should be placed on combining data from the existing studies as well as expanding autopsy-confirmed cohorts of these other neurodegenerative diseases.
Even though our study lacked power to detect disease risk association for these variants, we nonetheless leveraged the available in-depth neuropathology data to investigate their potential impact on specific neuropathologies. We first evaluated sub-categories of LBD-NP to determine whether any associations were driven by concurrent AD neuropathology, utilizing established criteria . We did not find statistically significant evidence for association of PLCG2_rs72824905-G with any of the three DLB sub-categories. In contrast, ABI3_rs616338-T was associated with increased risk of LBD-NP only in the intermediate category, which remained significant after adjusting for multiple testing. These findings may suggest that ABI3_rs616338-T associates with AD, but not LBD-NP, as the intermediate category has a higher burden of AD pathology unlike the high category. Alternatively, this variant may drive a specific combination of Lewy-related and AD pathology.
Our study also aimed to delineate the roles of these missense variants in influencing Aß and/or tau neuropathology as defined by Thal phase  or Braak stage , respectively. There was no association between ABI3_rs616338-T and either pathology. PLCG2_rs72824905-G associated with lower Braak scores and lower quantitative tau neuropathology, which is congruent with its protective effect seen in AD [8, 22, 24], some studies of DLB-CL [8, 24] and frontotemporal dementia (FTD) . Tau neuropathology is sine qua non in AD, frequently concurrent in LBD  and the main pathology in frontotemporal lobar degeneration due to tau (FTLD-tau), which constitutes about 40% of all FTLD . As tau is a common pathology in these conditions, it is plausible that a genetic variant that reduces risk across these diseases acts via suppressing tau neuropathology.
It is notable that the tau-suppressive effect of PLCG2_rs72824905-G is more pronounced in PSP Braak stage phenotype (beta = − 0.995, p = 0.01) than that for LBD (beta = − 0.292, p = 0.6). Braak scores are on average higher in our 971 LBD (mean = 3.9) than in 1036 PSP patients (mean = 2.5). This suggests that PLCG2_rs72824905-G may have a stronger suppressive effect on tau in earlier Braak stages.
Our results are aligned with those from a recent study of longitudinally followed clinical patients with mild cognitive impairment (MCI), where PLCG2_rs72824905-G was associated with lower cerebrospinal fluid (CSF) levels of pTau181 and cognitive decline . Collectively, our study which is focused on tau neuropathology and the published work on CSF tau  are consistent with a model where the effect of PLCG2_rs72824905-G in suppressing hyperphosphorylated tau may be most pronounced in the early stages of tau cortical deposition.
In addition to a suppressive effect as detected by Braak scores, there was also significant association of PLCG2_rs72824905-G and reduced quantitative neuropathology in PSP for different tau lesions, namely oligodendroglial coiled bodies (CB), tufted astrocytes (TA), tau threads (TAUTH) and neurofibrillary tangles (NFT). Despite a consistent effect on tau neuropathology in PSP, PLCG2_rs72824905-G does not have any effect on risk of this neurodegenerative disease. There may be several explanations for these results. Unlike AD, which has multiple neuropathologies, PSP is a primary tauopathy. It is possible that suppression of tau neuropathology by PLCG2_rs72824905-G may have a protective effect in neurodegeneration in the context of other neuropathologies but not when tau is the primary proteinopathy. Consistent with this possibility is the the suppressive effect of PLCG2_rs72824905-G on CSF pTau181, which was most pronounced in those MCI patients who also had evidence of Aß deposits based on low CSF Aß42 levels .
Collectively, these results may support the hypothesis that Aß and possibly also α-synuclein amplify the tau-suppressive effects of PLCG2_rs72824905-G in the early stages of tau cortical deposition. Using our human brain gene expression data [2, 3], we previously showed that PLCG2 resides in a brain co-expression network enriched for microglial genes and that brain PLCG2 levels were higher in AD but not in PSP compared to controls . We also discovered higher brain levels of plcg2 in two mouse models of amyloidosis compared to non-transgenic littermates . Recently, the protective PLCG2_rs72824905-G variant was shown to be a functional hypermorph, which increased the enzymatic activity of PLCγ2 . Our human and mouse gene expression results suggest that Aß but not tau leads to increased brain PLCG2 levels, either through microgliosis, microglial activation or both. Hence, the combination of increased levels of brain PLCG2, due to presence of Aß and higher enzymatic activity imposed by PLCG2_rs72824905-G may be necessary for an ultimately protective effect on disease risk. Though supported by multiple lines of evidence, this hypothesis requires further confirmation by testing the effects of PLCG2_rs72824905-G on neuropathology and other outcomes in model systems of different proteinopathies. To determine whether α-synuclein, like Aß, also leads to elevated brain levels of PLCG2, LBD cohorts without AD neuropathology need to be evaluated in transcriptome studies. Additionally, the influence of PLCG2_rs72824905-G on tau neuropathology and disease risk needs exploration in other conditions such as corticobasal degeneration (CBD) and FTLD-tau to determine the generalization of our results to other primary tauopathies.
Our study has numerous strengths, including assessment of two missense AD risk variants in two large autopsy-confirmed LBD-NP and PSP cohorts, investigating their effects on LBD-NP sub-categories representing different levels of LB and AD pathologies, detailed analyses of Aß and tau as endophenotypes including quantitative tau neuropathologies in 841 PSP patients. Despite these strengths, our study has some shortcomings. Even with our sizeable autopsy-confirmed cohorts, we were underpowered to detect associations given the low frequency of these variants. This raises the possibility of both false negative and false positive findings, although we note that the ABI3_rs616338-T association in the intermediate category would withstand correction for multiple testing. While we were able to assess both LBD-NP and PSP cohorts for genetic associations with Braak stage and Thal phase, quantitative tau measures were available only for the latter. Our study was conducted in self-reported Caucasian participants, which may not be a true representation of their genetic ancestry. Additionally, these results cannot be generalized to non-Caucasian populations. We determined previously that ABI3_rs616338-T and PLCG2_rs72824905-G are even rarer in African Americans than in Caucasians , suggesting that they have a smaller or no effect on AD risk in African Americans. Future studies should screen ABI3 and PLCG2 in sizable non-Caucasian cohorts to uncover the spectrum of genetic variants in these genes that may influence disease risk and neuropathology in non-Caucasian populations. Finally, both disease and neuropathologic associations should be investigated in other autopsy-confirmed neurodegenerative diseases to establish the generalizability of our conclusions to other conditions.
In summary, in our study of 973 LBD-NP and 1040 PSP autopsy-confirmed patients, we find evidence that the protective effect of PLCG2_rs72824905-G may be driven by suppressing tau especially in the earlier stages of its cortical deposition and may require presence of another proteinopathy, such as Aß to confer reduced disease risk. There is evidence of increased risk with ABI3_rs616338-T in a subset of LBD-NP patients with moderate to high AD pathology. These findings highlight potential mechanisms of action for these variants and exemplify utilization of detailed neuropathology phenotypes to untangle precise effects of genetic factors in these complex and heterogeneous neurodegenerative diseases.