In this study, we comprehensively investigated the association of liver function markers with AD diagnosis, amyloid PET burden, CSF biomarkers for AD, and cognition in two independent cohorts. A particularly novel aspect of this study was its emphasis on how this association might differ based on the presence of the APOE ε4 allele.
The findings shed light on significant associations between liver enzyme levels, particularly ALT levels and the AST to ALT ratio, with AD diagnosis, amyloid PET burden, and cognitive function, especially among individuals carrying the APOE ε4 allele. Notably, the AST to ALT ratio exhibited the strongest association in both the independent cohorts. However, these associations were not observed in those without the APOE ε4 allele. Furthermore, in the ADNI cohort, we observed that ALP levels were significantly associated with AD diagnosis, amyloid PET burden, and cognition, but this association was specific to the APOE ε4 carrier group. Additionally, the AST to ALT ratio was significantly associated with CSF Aβ42 levels in the ADNI cohort, especially in the APOE ε4 carrier group, but no such correlation was found for p-tau181 or t-tau levels. These findings strongly suggest that the presence of the APOE ε4 allele may play a crucial role in Aβ accumulation and cognitive decline in AD, potentially through its impact on liver function.
Intriguingly, our mediation analyses, conducted in two independent cohorts, demonstrated that age did not mediate the associations of liver enzymes with amyloid PET burden and AD diagnosis in the APOE ε4 carrier group. However, age played a mediating role in these relationships in the APOE ε4 non-carrier group. Importantly, the associations between liver enzymes and AD diagnosis were mediated by the amyloid PET burden, and this mediation effect was particularly observed in the APOE ε4 carrier group.
This study has provided insights into the influence of the APOE ε4 allele on the association of serum liver enzymes with AD pathogenesis and longitudinal changes in cognition, as evidenced by the analysis of two independent cohorts. Moreover, our findings also highlighted that the effect of liver enzymes on AD was mediated through age and amyloid PET burden, with the impact of this mediation varying depending on whether an individual carries the APOE ε4 allele.
The APOE gene encodes APOE, a 35-kDa glycoprotein with widespread expression throughout the human body that serves as a key lipid transporter [15]. Notably, the APOE ε4 allele is recognized as the most potent genetic risk factor for AD, and its influence increases in a gene dose-dependent manner [14]. In comparison with the APOE ε3 or ε2 allele, the APOE ε4 allele significantly increases the risk of AD by promoting the accumulation of Aβ in the brain [60]. Astrocytes are the primary source of APOE, which aids in the transportation of cholesterol to neurons via APOE receptors within the brain [16]. Conversely, in peripheral tissues, hepatocytes are the primary producers of APOE, which is released into the bloodstream to regulate cholesterol metabolism in an isoform-dependent manner [17]. APOE, primarily generated by the liver, is distinct from the form found in the brain and separated by the blood–brain barrier [17]. Despite their physical separation, mounting evidence suggests that peripheral APOE could potentially influence insulin signaling, neuroinflammation, and synaptic function in the brain [61, 62]. Moreover, plasma levels of APOE isoforms have been found to correlate with regional brain volume, cerebral glucose metabolism, and cognitive performance [63]. In mice models, ApoE4 has been shown to impede the Aβ peripheral clearance [18], while a biologically inspired nanostructure known as ApoE3-reconstituted high-density lipoprotein, which exhibits a strong binding affinity to Aβ, has been found to restore memory deficits by accelerating Aβ clearance [64]. Moreover, expression of ApoE4 in the liver has been found to exacerbate brain amyloid pathology, whereas liver-expressed ApoE3 has demonstrated beneficial effects on brain function and mitigated amyloid deposition in a mouse model [19]. Despite these intriguing associations, there is no consensus regarding the exact relationship between the APOE allele and AD pathogenesis. For instance, Huynh et al. reported that the deletion of ApoE in the hepatocytes of APP/PS1 mice, resulting in decreased plasma ApoE levels but no change in brain ApoE levels, did not influence the amount of amyloid plaques [65]. Furthermore, there is supporting evidence that serum-based liver function markers, including AST, ALT, and ALP levels and the AST to ALT ratio, are associated with AD diagnosis, poor cognitive performance, and increased Aβ deposition [6–11]. Additionally, there is a lack of studies investigating the effects of the APOE ε4 allele on the association of liver function markers with AD pathogenesis and cognition in humans. However, some of these concerns are partially addressed by the findings of our study, which showed that ALT levels and the AST to ALT ratio were significantly associated with AD diagnosis, Aβ accumulation, and cognition but only in the APOE ε4 carrier group across the two independent cohorts. Notably, such a correlation was not evident when assessing p-tau181 or t-tau levels. In particular, our mediation analysis revealed that the brain Aβ burden mediated the association between liver function markers and AD, exclusively in APOE ε4 carrier group. These findings indicate that the role of liver function in AD can be attributed to the accumulation of Aβ in the brain and that this relationship is dependent on an individual’s APOE ε4 carrier status.
The primary endogenous peripheral receptor responsible for regulating plasma Aβ, thereby preventing Aβ access to the brain, is circulating low-density lipoprotein receptor (LDL)-related protein 1 (LRP1) [66, 67]. In the liver, LRP1, in conjunction with LDL receptor, plays a crucial role in the clearance of circulating Aβ and APOE-containing particles from the bloodstream [67, 68]. The decrease in LRP-1 expression is implicated in age-related decline in hepatic Aβ clearance [66]. This impaired degradation of Aβ in the liver may lead to increased accumulation of Aβ in the brain [69]. In our mediation analyses results, we observed that the association of liver function markers with brain Aβ burden and AD was mediated by age, observed in the APOE ε4 non-carrier group. However, these mediation effects were absent in the APOE ε4 carrier group. These findings suggest that the age-related effect of liver function on Aβ-related AD pathogenesis is dependent on an individual’s APOE ε4 carrier status.
The implications of liver function in the pathogenesis of AD offer intriguing insights into the potential therapeutic targets for AD. For instance, the notable therapeutic effects of the ayurvedic agent, Withania somnifera, achieved by increasing levels of liver LRP, suggests that targeting peripheral Aβ clearance may provide a unique approach to rapidly eliminate Aβ in AD transgenic mice [70]. Statins have also demonstrated the potential to reduce the risk of AD by upregulating hepatic LRP1 and LDL receptor expression, which is mediated by sterol response element-binding protein-2 [71, 72]. Additionally, transthyretin, a transporter protein primarily produced in the liver and released into the bloodstream, is downregulated in AD [73]. Given its role as a carrier of Aβ at the blood–brain barrier and in the liver, particularly via LRP1, transthyretin may offer valuable insights into the development of therapeutic strategies for AD [74, 75].
One notable limitation of our study was the inability to incorporate quantitative measures of amyloid PET into the HUMC cohort when investigating the relationship between liver enzymes and the amyloid PET burden. This limitation arises from the absence of imaging data, which restricts the depth of the analysis. However, the practice of visually rating amyloid PET scans remains valuable in the clinical setting. Second, the HUNC cohort did not have data on BMI, which is an important covariate associated with ALT levels [76]. The absence of BMI data could have affected our results and their interpretations. Third, the HUMC cohort did not have data on the CSF biomarkers for AD. This limitation restricted our ability to explore associations with CSF neuropathology, and our investigation in this regard was confined to the ADNI cohort. Fourth, we generated composite scores for the four distinct cognitive domains using different individual cognitive tests in the two cohorts. This variation in cognitive assessments may have introduced some degree of variability to our results. Finally, differences in the frequency and definitions of the diagnostic groups between the two cohorts could have potentially introduced confounding factors into our study.