Association Between Serum Ceruloplasmin Specific Activity and Risk of Alzheimer’s Disease

INTRODUCTION

A solid body of literature gathered over the last 10 years has shown that dysfunction in human metal metabolism, primarily involving copper, is associated with the sporadic form of Alzheimer’s disease (AD) [1, 2]. In vitro models of copper toxicity [3–5], along with evidence from experimental animal models, showed that copper enhances the toxicity of amyloid-β (reviewed in [1, 6]). Clinical studies have shown some systemic abnormalities in copper metabolism in relation to ATP7B gene variants [7]; these abnormalities are shared between AD and Wilson’s disease (WD), the paradigmatic disease of copper toxicosis or accumulation [6, 9].

Increasing evidence helps define abnormalities affecting the pathway of copper regulation at the hepatic level. This pathway normally regulates the excretion of copper and the oxidation of systemic iron and encompasses serum total copper, non-ceruloplasmin bound copper (Non-Cp Cu, also known as “free” copper, or labile copper), ATP7B protein, and ceruloplasmin (Cp). Although much less severe than in WD, the increase in Non-Cp Cu correlates with some typical AD deficits, and with the ‘core’ markers of the AD in the cerebrospinal fluid [10]. A poor prognosis of the disease [11], along with increases in the rate of the conversion from mild cognitive impairment to full dementia [12], has also been described. Meta-analyses have confirmed increased levels of copper [13–15], and of Non-Cp Cu in sera of AD patients [16], and decreased concentrations of total copper in the brains of AD patients [15], in line with a systemic Non-Cp Cu imbalance [6]. A recent meta-analysis evidenced a decreased concentration of zinc levels in the serum, plasma, and cerebrospinal fluid of AD patients [17]. The involvement of zinc alterations in AD and their interplay with copper metabolism [18] suggests the employment of zinc therapy with respect to other copper-chelating agents [19]. Our previous studies have shown an association between ATP7B gene variants and AD risk [7, 21], supporting a potential causative role of Non-Cp Cu in the disease pathogenesis since the early and pre-clinicalstages.

Cp is a serum protein that is tested to evaluate systemic copper status, due to its tight relationship with the ATP7B protein. In WD, ATP7B mutations modify copper loading into nascent Cp, producing an apo-Cp form. This apo-Cp is partly secreted into plasma and rapidly degraded into fragments that lack enzymatic activity [22, 23]. The commercial assays employed to test serum Cp concentration do not discriminate between apo-Cp and the completely active form of the protein: the holo-Cp. By analyzing Cp activity, we can integrate the commercial assay for Cp concentration, which is one of the markers of WD (values <  20 mg/dL), reported in the Leipzig scoring system [24]. In fact, patients with apparently normal values of Cp concentrations could have lesser values of Cp activity and could be erroneously excluded from further testing for WD, particularly when other copper assays are borderline [22]. The same concept can be applied to Non-Cp Cu dismetabolism in AD. On this basis, we have decided to further investigate the role of Cp in AD, considering the heterogeneous results for Cp concentration and the limited information on Cp activity in AD—as reported in previous studies and in a recent meta-analysis [15]. Herein, we suggested Cp specific activity (eCp/iCp) as a more precise biochemical variable to test Cp, examining its association with the risk of AD, along with Non-Cp Cu and other metal markers in recently published data.

MATERIALS AND METHODS

RESULTS

The demographic, biochemical and genetic characteristics of AD patients and healthy controls are shown in Table 1.

The univariate logistic analyses revealed, as expected, significant associations of age, gender, and APOE-ɛ4 allele (APOE-ɛ4; carriers of at least one allele versus ɛ4 non carriers) with the probability of having AD. The associations of Non-Cp Cu, eCp/iCp, Cu, iCp, Tf, and Cp/Tf with AD were also significant (Table 2).

In the multivariable analysis, gender did not significantly impact the probability of having AD (p = 0.061); therefore, only age was used for correction purposes. The variables considered were: APOE-ɛ4, eCp/iCp, Fe, Non-Cp-Cu, and Cp/Tf ratio. To avoid multi collinearity, we excluded biochemical variables that were deemed “redundant”, such as Cu, iCp, eCp, and Tf. The forward stepwise procedure identified as the best biochemical and genetic subset of predictors of diagnosis of AD: APOE-ɛ4, eCp/iCp, and Non-Cp Cu (Table 2).

Possible interactions between age and the selected biochemical variables were evaluated, but they were not significant so we did not include them into the final model.

The estimated model confirmed the role of age as risk factor for AD: an increase of one year of age raised the probability of having AD by 12% (adjusted OR = 1.12, 95% CI = 1.06–1.18; p = 0.001), and revealed a potential protective effect of eCp/iCp. That is, with an increase of one unit in the eCp/iCp values, the adjusted odds of having AD were reduced by 89% (adjusted OR = 0.11, 95% CI = 0.03–0.39; p = 0.001). Conversely, increases of one μmol/L unit of Non-Cp Cu levels raised the adjusted odds of having AD by 60% (adjusted OR = 1.60, 95% CI = 1.13–2.26; p = 0.008) (Table 2). For a subject carrying an APOE-ɛ4 allele, the adjusted probability of having AD was 10.18 higher with respect to that of a subject non-APOE-ɛ4 carrier (adjusted OR = 10.18, 95% CI = 2.92–35.56). Tests for multicollinearity indicated that a very low level of multicollinearity was present (VIF = 1.08 for carriers of at least one APOE-ɛ4, 1.22 for age, and 1.60 for eCp/iCp, 1.55 for Non-Cp-Cu). The estimated model appeared to fit data quite well (Hosmer and Lemeshow test: χ² (8)  = 9.072; p = 0.336) and showed a ‘good’ capacity to discriminate AD patients from healthy controls (AUC = 88% , 95% CI = 83–94%) (Fig. 1). To derive the ‘optimal’ cut-off for sensitivity and specificity for the estimated model, as well as for the significant predictors, Youden’s index was calculated and maximized. As Table 3 shows, the model reached the highest sensitivity with a cut-off of 0.75, with respect to those of the variables evaluated separately. The exact McNemar test was performed to compare the sensitivity and specificity of the predictors; it revealed that the sensibility of the estimated model was higher than those of the eCp/iCp and Non-Cp Cu evaluated separately (both p values = 0.020). The specificity of the estimated model was higher than of the Non-Cp Cu one only (p = 0.004).

DISCUSSION

The main result of this study is that eCp/iCp is associated with a decreased risk of developing AD. Conversely, as has been depicted in previous studies, Non-Cp Cu in this new subject sample increases the risk of having AD [11, 36–38]. In other words, eCp/iCp emerges as a potential protective factor that contrasts the effects of an altered Non-Cp Cu. More specifically, a one-unit increase of eCp/iCp decreased the probability of having AD by 89% (OR = 0.11, p = 0.001). Conversely, a one-unit increase of Non-Cp Cu increased the probability of having AD by 60% (OR = 1.60, p = 0.008) (Table 2).

As expected, subjects of advanced ages were shown to have a higher probability of contracting AD. Specifically, an increase of one year of age heightened the risk of having AD by 12% (adjusted OR = 1.12, p = 0.001). The effect of age was always taken into account in the multivariable analysis, and none of the interactions between age and the biological variables included in the final model was significant. In addition, there was no significant correlation between age and Non-Cp Cu (Spearman’s correlation coefficient (rs) = 0.244, p = 0.065 in healthy controls; rs = 0.015, p = 0.894 in AD) and eCp/iCp (rs = –0.200, p = 0.132 in healthy controls; rs = –0.113, p = 0.306 in AD). As a result, the effects of Non-Cp Cu and eCp/iCp could not be ascribed to differences in age between AD subjects and controls.

The univariate logistic analyses showed that, apart from Fe and eCp, all the biochemical variables were associated with AD. This finding confirms previous conclusions reached about copper and iron derangement in AD. In fact, Cu values increase and correlate with the risk of AD (reviewed in: [39]); otherwise Tf values decrease and Cp/Tf ratio values increase in association with the probability of the disease [40, 41].

The final multivariable forward stepwise logistic estimated model includes age, APOE, eCp/iCp, and Non-Cp Cu. This model has a good power in discriminating AD patients from healthy controls (AUC = 88%) with a sensitivity of 66% and a very high specificity, of 93% (Table 3; Fig. 1). In the model, APOE alone has a strong effect (adjusted OR = 10.18, 95% CI = 2.92–35.56; p <  0.001). It shows a sensitivity of 44% and a specificity of 91% ; this could be caused by the APOE-ɛ4 allelic distribution of the sample. In fact, only 8.6% of healthy volunteers were carriers of at least one ɛ4 allele. Conversely, 44% of the AD subjects were carriers (Table 1). This allelic distribution differs from the percentages reported in the literature for Caucasians (14% in healthy and 38% in AD) [42].

The high specificity of the model could also be ascribed to copper status variations, as Non-Cp Cu and eCp/iCp encompasses 72% and 83% of the healthy controls, respectively (Table 3). A literature overview of Cp measurements in AD could provide the framework needed to interpret these results. In fact, a recent meta-analysis reported no difference in AD Cp concentration (iCp), or in Cp activity (eCp) [15]. In that meta-analysis, however, two studies published recently [43, 44] were not included. As a whole, the published data [10, 43–58] indicate that iCp values in AD are heterogeneous and discordant. More specifically, fifteen of the nineteen studies cited show no difference in iCp values [10, 48–58]. In one study, the iCp values were lower in AD subjects than in healthy controls [45]; in three studies, the iCp values were higher in AD subjects than in healthy controls [43, 47]. Among these nineteen studies, only three measured the eCp [53, 56], which was always lower in AD subjects, as confirmed by the meta-analysis [15]. Recently, one additional study reported that the values of total serum ferroxidase activity—ascribable principally to Cp—were elevated in AD, even though they did not reach the statistical threshold [59].

Regarding the eCp/iCp variable, we considered only studies that provided complete information on the number of patients and the mean and SD or median and min-max. As a result, we considered only two studies; both reported that values in AD patients were lower than in healthy controls [53, 60]. Considering our results—and those reported in the literature—decrease of eCp/iCp is evident. It suggests, in AD, an impairment in the biosynthesis of the totally active holo-Cp with a failure in copper incorporation or a premature loss of copper. In fact, a shift in the balance between the holo- and apo-form of the protein toward that of apo-Cp (i.e., toward less eCp/iCp) can be posited. Current results sustain previous findings of increased levels of apo-Cp in the serum of AD patients [61, 62], along with signs of liver hypo-function ascribable to disturbing effects of Non-Cp copper on hepatocytes[63].

The ratio eCp/iCp was proposed for the first time by Milne and Johnson [64] as a helpful indicator of copper systemic status. That study measured Cp concentration and activity, copper and the ratio eCp/iCp in a population of 141 healthy volunteers, divided across six decades. It found that Cp concentration and copper were affected by age, gender, and hormone use, while the ratio eCp/iCp was not [65], suggesting that this variable is a reliable marker of copper status. eCp/iCp was then employed to assess copper status in pregnant women, newborns and epileptic patients [66–68]. Herein the eCp/iCp was rapidly estimated, thanks to an automation of the eCp assay [33], which simultaneously measures the activity and the immunologic concentration of Cp through an automatic biochemical analyzer. This procedure can be employed easily in clinical setting to analyze copper disorders; we demonstrated that eCp/iCp differentiated between different liver diseases (WD, hepatic encephalopathy, and other chronic liver disorders) better than iCp or eCp alone [33]. In fact, subjects with lower eCp/iCp values may have a “defective” Cp, with a Cp concentration that is higher than the value of enzymatic activity expected from the fully active protein [33]. This is consistent with the finding that, in WD, not every mutation of the ATP7B-gene lowers levels of iCp, and in some cases of later onset WD an apparently normal iCp value may hide a copper imbalance [69]. Moreover, recent works reported a post-translational modification of Cp due to oxidative damage, which could limit the activity of Cp [70]. Taking this into consideration, iCp or eCp should not be analyzed alone. Rather eCp/iCp should be tested to evaluate systemic copper status in AD as well as in WD. Moreover, in the event of a Non-Cp Cu alteration, during anti-copper therapy (particularly zinc therapy), the eCp/iCp should be measured to monitor the Cp status to prevent adverse effects caused by overtreatment.

The preliminary study on this sample suggests that the eCp/iCp ratio is a useful marker of Cp systemic status that should be tested to support investigation into copper homeostasis in AD. However, further investigation on a larger sample is necessary to confirm this result and the employment of this variable in clinical diagnosis.