RESEARCH ARTICLE

Value of Serum Nonceruloplasmin Copper for Prediction of Mild Cognitive Impairment Conversion to Alzheimer Disease Rosanna Squitti, PhD,1,2 Roberta Ghidoni, PhD,3 Mariacristina Siotto, PhD,4 Mariacarla Ventriglia, PhD,1 Luisa Benussi, PhD,5 Anna Paterlini, PhD,3 Mariachiara Magri, PhD,6 Giuliano Binetti, MD,5 Emanuele Cassetta, MD,1 Deborah Caprara, PhD,1 Fabrizio Vernieri, MD,7 Paolo M. Rossini, MD,8,9 and Patrizio Pasqualetti, PhD6,10 Objective: Meta-analyses show that nonbound ceruloplasmin (non-Cp) copper (also known as free or labile copper) in serum is higher in patients with Alzheimer disease (AD). It differentiates subjects with mild cognitive impairment (MCI) from healthy controls. However, a longitudinal study on an MCI cohort has not yet been performed to assess the accuracy of non-Cp copper for the prediction of conversion from MCI to AD during a long-term follow-up. Methods: The study included 42 MCI converters and 99 stable MCI subjects. We assessed levels of copper, ceruloplasmin, non-Cp copper, iron, transferrin, ferritin, and APOE genotype. A multiple Cox regression analysis—with age, sex, baseline Mini-Mental State Examination, APOE4, iron, non-Cp copper, transferrin, ferritin, hypercholesterolemia, and hypertension as covariates—was applied to predict the conversion from MCI to AD. Results: Among the evaluated parameters, the only significant predictor of conversion to AD was non-Cp copper (hazard ratio 5 1.23, 95% confidence interval 5 1.03–1.47, p 5 0.022); for each additional micromole per liter unit (lmol/l) of non-Cp copper, the hazard increased by 20%. Subjects with non-Cp copper levels >1.6lmol/l had a hazard conversion rate (50% of conversion in 4 years) that was 33 higher than those with values 1.6lmol/l (13; (5) other psychiatric diseases, epilepsy, drug addiction, alcohol dependence, and use of psychoactive drugs, including acetylcholinesterase inhibitors or other drugs that enhance cognitive functions; and (6) current or previous uncontrolled or complicated systemic diseases (including diabetes mellitus) or traumatic brain injuries.11 Additional exclusion criteria were conditions known to affect copper metabolism and biological variables of oxidative stress on the basis of past medical history and screening laboratory tests; this is reported in detail elsewhere.21 We also excluded patients with a history of stroke, presence of focal neurological signs, ischemic lesions or severe corticosubcortical leukoencephalopathy on brain magnetic resonance scans, presence of atherosclerotic plaques and hemodynamically significant stenosis or occlusion in cerebral (neck and intracranial) arteries, or history of coronary artery disease (myocardial infarction, angina, coronary artery angioplasty, or bypass surgery) or cardiac arrhythmias (atrial flutter or fibrillation) and history of peripheral arteriopathy. Blood was drawn on the same day of the clinical interview.

Biochemical and Molecular Investigations Patients had samples drawn at their baseline visit. Fasting blood samples were collected in the morning and the serum was quickly stored at 280 C for blind biochemical measurements. We measured copper by spectrometry,11 using an A Analyst 600 PerkinElmer (Waltham, MA) atomic absorption spectrophotometer. We analyzed the ceruloplasmin with an immunoturbidimetry assay (Horiba ABX, Montpellier, France). To better circumstantiate the immunoturbidimetry assay used for non-Cp copper calculation, we determined the serum ceruloplasmin oxidase activity with o-diansidine dihydrochloride as a substrate spectrophotometrically. The methodological details are reported in a previous study.22 For the non-Cp copper calculation, we used values of ceruloplasmin obtained immunologically.22 More precisely, for each serum copper and ceruloplasmin pair, we computed the amount of copper bound to ceruloplasmin (CB) and the amount of non-Cp copper, following standard procedures.12 Briefly: CB 5 ceruloplasmin (mg/dl) 3 10 3 n; n 5 0.0472 (lmol/mg); Non-Cp copper 5 absolute serum copper 2 CB.12 This calculation expresses non-Cp copper in lmol/l and is

575

ANNALS

of Neurology

based on the finding that ceruloplasmin binds 6 atoms of structural copper, which is equivalent to ceruloplasmin containing 0.3% copper (weight percentage of copper in Cp: [6 (atoms of Cu) 3 63 mass atomic units (Cu molecular weight)/132kDa (Cp molecular weight)] 3 100 5 0.3%; more details are available at http://www.j-alz.com/letterseditor/index.html# March2013). Thus, for a patient with a serum absolute (or total) copper concentration of 17.3lmol/l and a serum ceruloplasmin concentration of 33mg/dl, the bound copper concentration equals 33 3 10 3 0.0472 5 15.6lmol/l, and the non-Cp copper concentration is 17.3 2 15.6 5 1.7lmol/l.12 Measurements of the other biological variables of metals are described in detail elsewhere.10 We used an immunoturbidimetry assay to analyze transferrin and ferritin levels, and ferene to analyze iron. All of these reagents were from the ABX Pentra series from Horiba ABX. All biochemical measures were automated on a Cobas Mira Plus (Horiba ABX) and performed in duplicate. APOE genotyping was performed according to standard methods.23 Among the study population, 13 cases were not analyzed for APOE, because it was not possible to assess the genotype (due to insufficient DNA, n 5 8; insufficient blood samples, n 5 2; and polymerase chain reaction failure, n 5 1).

Data Analysis We started to recruit MCI patients in 2005 and followed up with them through 2013, observing an attrition rate of 10% per year. The minimum follow-up was 2 years, whereas the patient with the longest follow-up was observed for 6 years. To recognize as significant (at a 2-sided alpha level of 0.05) a hazard ratio (HR) of 2 between those patients with non-Cp copper >1.6lmol/l and those with non-Cp copper 1.6lmol/l (where the annual hazard rate was assumed to be 5%), we planned to recruit at least 140 MCI subjects to reach an adequate power (0.8). We further assumed that the whole sample was equally distributed below and above 1.6lmol/l. Because the aim of the study was to see whether and how the incidence of conversion to AD is related to non-Cp copper, we applied a multiple Cox regression analysis, taking into account demographic, clinical, and metal metabolism measures. To control the risk of overfitting, we limited the number of candidate predictors performing different regression analyses with different subsets. More precisely, to avoid multicollinearity, we refrained from entering correlated variables into the same subset (eg, because copper, ceruloplasmin, and non-Cp copper are linked through a linear function, we considered copper and ceruloplasmin to belong to one subset and non-Cp copper to another). In this way, the event per variable (EPV) ratio was maintained at >5. Nevertheless, EPV was not high and remained below the suggested rule of thumb of 10. Thus, the stability of our prognostic models could not be considered optimal. Risk proportionality was checked through the conventional methods (dependency of partial residuals, log 2 log plots). The number of events different from conversion to AD and censoring was very low (5 patients had a diagnosis of vas-

576

TABLE 1. Baseline Characteristics, MMSE Scores, and Metal Studies in the 141 Subjects with Mild Cognitive Impairment Participating in the Study

Characteristic

Value

Age, yr, mean (SD)

70.8 (8.1)

Sex, No. women [%]

66 [47]

Risk factors, % Hypertension

23

Hyperlipidemia

20

APOE e4 frequency, No. [%]

43/128 [33.6]

Biological variables in serum at baseline, mean (SD) Copper, lmol/l

14.8 (2.9)

Cp, mg/dl

27.7 (5.2)

Cp activity, IU/l

105.2 (21.7)

Non-Cp copper, lmol/l

1.8 (2.2)

Iron, lg/dl

100.0 (63.3)

Tf, g/l

2.5 (0.7)

Ferritin, ng/ml

141.2 (145.9)

Cp/Tf ratio

12.4 (6.7)

Cognitive evaluation MMSE score, mean (SD)

27.2 (1.5)

Cp 5 ceruloplasmin; MMSE 5 Mini-Mental State Examination; SD 5 standard deviation; Tf 5 transferrin.

cular dementia during follow-up, and they were not considered in the analysis). Therefore, no attempt to apply a competing risk model was made. To provide additional information potentially useful for clinicians, we entered non-Cp copper both as a continuous and as a dichotomous variable (predefined cutpoint 5 1.6 lmol/l).

Results Subjects’ Description The main baseline demographic, biochemical, and clinical features of subjects enrolled in the study are reported in the Table. The frequency of allele e4 of the APOE gene (APOE4) was 33.6%. Seventy-eight subjects (55%) had non-Cp copper levels higher than normal (>1.6 lmol/l).12 Prediction of Cognitive Decline Forty-two of 141 subjects (30%) converted to AD during the follow-up observational period. According to a Cox regression model—with sex, age, baseline MMSE, APOE4, iron, non-Cp copper, transferrin, ferritin, Volume 75, No. 4

Squitti et al: Copper and MCI Conversion

hypercholesterolemia, and hypertension as covariates— (sex, age, and baseline MMSE were entered as forced terms to adjust for their association), the only significant predictor of conversion to AD was non-Cp copper (HR 5 1.23, 95% confidence interval [CI] 5 1.03–1.47, p 5 0.022). For each additional lmol/l unit of non-Cp copper, the hazard increased by 20%. None of the other variables entered in the model had probability values consistently >0.15. Because non-Cp copper is a derived measure, we tested the model by removing non-Cp copper and entering as candidate predictors both components (copper and Cp). No variable passed the entry criteria (p < 0.05). When non-Cp copper was considered as a binary variable (with respect to the cutoff of 1.6lmol/l), its association was still significant (HR 5 3.34, 95% CI 5 1.21–9.24, p 5 0.020). This indicated that the hazard rate of MCI subjects with non-Cp copper >1.6lmol/l was 33 higher than of those with values 1.6lmol/l. As shown in the Figure, 1 of 5 subjects with high non-Cp copper values converted to AD within 1.5 years of baseline evaluation (compared to >4 years for patients with low non-Cp copper). That is, the first 20% of conversion occurred 2.5 years earlier in MCI subjects with high non-Cp copper. In addition, the median conversion time was 4 years in the first group and was not estimable in the second. To compare APOE4 with non-Cp copper in predicting MCI conversion to AD, we classified our patients as carriers (of at least 1 copy) or noncarriers of the APOE4 genotype. Among the 128 subjects analyzed, the rate of conversion was similar between APOE4 carriers and noncarriers (HR 5 1.4, 95% CI 5 0.7–2.8, p 5 0.321). The non-Cp copper association with AD conversion was independent with respect to the APOE4 association, as indicated by the lack of interaction (p 5 0.989), and by the confirmation of the significant association of non-Cp copper (p 5 0.036) when APOE4 was entered as a stratification variable. The lack of a correlation between partial residuals and time (r 5 0.002, p 5 0.990) and the linearity and parallelism of a log 2 log plot indicated that the proportionality assumptions were not violated.

Discussion The most important result of this study is that a deregulation of non-Cp copper has an association with the rate of clinical conversion from MCI toward AD full dementia in an MCI cohort that received clinical follow-ups for 6 years. Our current findings show that higher levels of non-Cp copper, in the very early stages of cognitive complains, could account for a faster rate of conversion to AD. This is in line with a previous study.14 Our data April 2014

FIGURE 1: Probability of remaining at mild cognitive impairment (MCI) status during the follow-up, according to nonbound ceruloplasmin (non-Cp) copper levels (open circles, £ 1.6; closed circles, >1.6). Half of the patients in the group with high non-Cp copper values converted to Alzheimer disease (AD) after approximately 4 years. At approximately the same interval,

Value of serum nonceruloplasmin copper for prediction of mild cognitive impairment conversion to Alzheimer disease.

Meta-analyses show that nonbound ceruloplasmin (non-Cp) copper (also known as free or labile copper) in serum is higher in patients with Alzheimer dis...
120KB Sizes 3 Downloads 4 Views