Plasma prion protein concentration and progression of Alzheimer disease

Abstract

Background/Objective: Recently, PrP^c has been linked to AD pathogenesis. Second, a relation of PrP^c plasma levels with cognitive status and decline of healthy elderly subjects has been reported. Therefore, we hypothesized baseline plasma levels of PrP^c to be associated with AD progression in cognitive and functional domains.

Materials and Methods: AD patients (n = 84) were included into an observational study at time of diagnosis. Baseline plasma PrP^c levels were determined. Decline was assessed annually (mean follow-up time 3 years) with the aid of different standardized tests (MMSE, iADL, bADL, GDS, UPDRSIII). Multiple regression analyses were used to uncover potential associations between decline and PrP^c levels.

Results: No association of PrP^c and decline could be established. Presence of diabetes mellitus was linked to slower deterioration. Intake of neuroleptic drugs or memantine was associated with faster progression.

Conclusion: Plasma PrP^c at baseline could not be shown to be related to AD progression in this study. An interesting association of diabetes mellitus and decline warrants further investigation.

Keywords: :

• Alzheimer
• prion protein
• cognition
• biomarker

Introduction

Alzheimer disease (AD) as the principal neurodegenerative condition leading to dementia is one of the major health problems today and even more so in the decades to come. With demographic change its momentum will increase dramatically.¹ In most parts AD pathogenesis is not fully understood. Thus, unsurprisingly, scientific efforts to shed light on causal pathomechanisms and to perform therapeutic trials have increased noticeably. Biomarker research plays a central role, since biomarkers might support earlier diagnoses and therefore potentially allow for early/earlier interventions. Biomarkers might serve as biochemical surrogates for disease activity and could therefore be of key interest in treatment trials as outcome parameters (besides neuropsychology) or even as therapeutic targets. This is especially true for preclinical AD, where measureable cognitive deficits are—by definition—yet absent.

One of the biomarkers of interest is the physiological form of the Prion protein (PrP^c).² PrP^c is a protein, which is expressed almost ubiquitously, with highest concentrations in the central nervous system.³ Its physiological functions are still enigmatic to some extent. Yet, it has been demonstrated, that PrP^c might be involved in inflammatory processes, protection from oxidative stress, cell proliferation, cell differentiation, apoptotic pathways, and signaling.^4,5 It has even been linked to learning processes in experiments with rodents.^6,7 In other animal studies, neuroprotective properties have been ascribed to PrP^c.^6,8,9 In its misfolded conformation (PrP^sc), the prion protein is known to be the key pathogenic agent causing various prion diseases such as Creutzfeldt-Jakob disease.¹⁰

PrP^c appears to have a role in AD pathophysiology. However, this is still matter of debate. It may act as a β-secretase regulator thereby affecting Aβ production.¹¹ There is also evidence, that it may be a mediator of Aβ oligomer toxicity with impact on long-term potentiation (LTP), synaptic and neuronal plasticity.^12-15 Alier et al. see a “latent pro-pathogenic role (of PrP^c) in the presence of amyloid assemblies”.¹⁶ It was possible to counteract harmful effects of Aβ and even restore parts of cognition using antibodies against PrP^c in animal studies.^17-20

Regarding PrP^c’s role as a potential biomarker, a recent study revealed serum PrP^c concentrations to be associated with cognitive status in a healthy, elderly population.²¹ Compared with healthy controls, AD patients feature lower PrP^c concentrations in the CSF (as is the case in other neurodegenerative conditions).²² Moreover, the Codon 129 polymorphism of the very gene encoding the prion protein (PRNP gene), which is known to affect prion disease phenotypes, has also been linked to cognition and speed of decline in AD.²³

Considering the evidence partially relating PrP^c to AD, we hypothesized, that plasma concentrations of PrP^c might be associated with the velocity of decline in AD. Thus, we herein analyzed a group of 84 patients in this respect.

Results

The sample size was 84 patients. As to be expected, there was a slight shift toward female gender (57%). Mean age at diagnosis was 66 y. Mean baseline MMSE status was 20pts. Complete medical, socio-demographic/socio-biographic, genetic and neuropsychological data are given in Table 1. Mean follow-up time was 3 y (SD 1 y). All relevant CSF biomarkers were altered as necessary for the diagnosis of AD (due to skew distribution values are given as: median (IQR)): Aβ42 = 422(288) (laboratory specific cut-off 450) [pg/mL]; Aβ40 = 6803(4648) (cut-off 450) [pg/mL]; Aβ-Ratio: 0.6(0.3) (cut-off 1); Tau: 488(538) (cut-off 450) [pg/mL]; pTau: 86(65) (cut-off 61) [pg/mL]). Median plasma PrP^c concentration at baseline was 0.88 (IQR = 0.66) [ng/mL].

Table 1. Sociodemographic, medical, genetic and neuropsychological data

Item		Value	n
age	mean (SD) [yrs]	66.1 (9.65)	84
gender	m:f [n(%])	36:48 (43:57)	84
years of formal education	< 8yrs, 8yrs, 8–10yrs, 11–12yrs, > 12yrs [%]	8, 49, 27, 2, 13	84
presence of diabetes	%	17	84
presence of hypertension	%	50	84
APOE genotype	2/2, 2/3, 2/4, 3/3, 3/4, 4/4 [%]	2.6, 2.6, 3.9, 43.4, 36.8, 10.5	84
PRNP genotype	M/M, M/V, V/V [%]	34, 46, 20	76
drug intake, AChEI	fraction of observation time, mean (SD)	0.568 (0.3843)	84
drug intake, memantine	fraction of observation time, mean (SD)	0.293 (0.3870)	84
drug intake, neuroleptics	fraction of observation time, mean (SD)	0.266 (0.3478)	84
MMSE baseline	mean (SD) [pts]	19.93 (6.912)	84
MMSE progression/year	mean (SD) [pts/yr]	-2.54 (3.253)	84
bADL baseline	mean (SD) [pts]	43.40 (5.685)	84
bADL progression/year	mean (SD) [pts/yr]	-3.49(3.649)	82
iADL baseline	mean (SD) [pts]	5.25(2.620)	84
iADL progression/year	mean (SD) [pts/yr]	-0.93 (1.009)	82
GDS baseline	mean (SD) [pts]	3.70 (0.991)	84
GDS progression/year	mean (SD) [pts/yr]	0.31 (0.519)	82
UPDRSIII baseline	mean (SD) [pts]	12.95 (14.849)	83
UPDRSIII progression/year	mean (SD) [pts/yr]	3.16 (6.913))	74

In the multivariate analyses no association of PrP^c concentration could be established in any of the domains considered, although the effect on UPDRSIII came close to statistical significance (P = 0.053). Presence of diabetes mellitus was associated with slower decline in some domains, while intake of memantine or neuroleptic drugs were associated with faster progression. Higher baseline iADL score could be linked to faster instrumental ADL decline, while worse UPDRSIII or GDS scores were related to slower progress on these respective scales. Detailed data are given in Table 2.

Table 2. Multivariate analyses

	MMSE		bADL		iADL		GDS		UPDRSIII
	Est.	P	Est.	P	Est.	P	Est.	P	Est.	P
PrP^c (log)	0.1536	0.6006	0.2232	0.4778	-0.0315	0.7169	-0.0186	0.6394	-1.1848	0.0528
sex (f vs. m)	-0.0812	0.9166	0.8806	0.2641	0.034	0.8747	-0.0945	0.3473	0.8645	0.5904
age	0.0017	0.963	-0.0352	0.4002	-0.0003	0.9792	0.0018	0.7269	-0.0455	0.5794
education	-0.223	0.4744	0.1619	0.6405	-0.1885	0.05	0.042	0.335	0.7579	0.2554
baseline	-0.082	0.1579	-0.0262	0.7262	-0.1831	0	-0.202	0.0004	-0.1586	0.0157
neuroleptics	n.a.	n.a.	-2.8309	0.0236	-0.9473	0.0048	0.7118	0	10.1785	0.0002
memantine	-1.9949	0.0346	-2.6934	0.0088	n.a.	n.a.	0.6467	0	n.a.	n.a.
diabetes	3.3204	0.0006	3.4358	0.0012	n.a.	n.a.	-0.4773	0.0004	-3.245	0.1295

Est. (estimate) gives the difference of the slopes between groups for binary data. For continuous data, it resembles a regression coefficient. For logarithmic data [PrP^c (log)] it can be interpreted as follows: for every 1% change of the value, the slope changes by a hundredth of the estimate. Significant results (P < 0,05) are given in bold font face. (n.a., not available, not included into the final model during the modeling process).

Discussion

Rationale for performing this study was scientific evidence indicating PrP^c to be involved in AD pathogenesis (as described above). Since PrP^c can reliably be measured in plasma of patients and has already been associated with cognitive function in humans, it stood to reason to investigate PrP^c as a potential biomarker in AD.

As a main result, we were not able to demonstrate a statistically significant association of PrP^c plasma levels at baseline with cognitive, functional or motor decline in AD. Possible explanations might be, that, first, PrP^c levels are either generally not suitable as a biomarker because there is indeed no real correlation or second, PrP^c plasma levels do not sufficiently reflect central nervous processes. Possibly, measurement of PrP^c in the CSF might be more informative. But likewise, even then it is not clear, whether CSF concentrations correlate to AD specific, pathogenetic processes within the brain at all. Although shown, that differential CSF PrP^c levels occur in health and various neurodegenerative conditions, associations of CSF PrP^c to cognitive status have not been demonstrated.^22,24 Regarding peripherally measurable PrP^c, it has to be considered furthermore, that PrP^c is not only expressed in central nervous structures but also in peripheral tissues.³ PrP^c is a relatively small molecule and has been thought to easily cross the blood brain barrier, but additionally, recent studies provide evidence of bidirectional transcytotic mechanisms.²⁵ Considering all these factors, which might have a relevant impact on PrP^c plasma levels, the results of Breitling et al. appear rather intriguing as they revealed associations of cognition with PrP^c plasma levels.²¹ As it appears not feasible to associate baseline PrP^c values with future deterioration, it, hypothetically, might still be conceivable, that intra-individual differences in serial PrP^c measurements over the disease course might reflect disease progression. A currently ongoing study will address this question.

Some secondary findings need to be discussed as well. Intake of neuroleptic drugs and the duration of drug intake correspond well to faster decline in all domains. This is in line with the published literature.²⁶ Intake of memantine as an anti-dementia drug is associated with faster deterioration in this study. This must not be interpreted as causal. Although treatment guidelines exist, in our experience from everyday clinical routine many physicians are inclined to deviate from guidelines: patients who initially decline faster might be more likely to receive memantine as an early off-label add-on medication, because beneficial effects may be assumed.²⁷ Surprisingly, the presence of diabetes mellitus as a comorbidity appears partially protective of fast decline. Contradicting results have been published concerning this matter.^28-30 AD and diabetes seem to share some common pathogenetic pathways. Thus, trials using anti-diabetic drugs to achieve AD disease modulation have been performed.³¹ Conceivably, our findings could be confounded by the intake of anti-diabetic drugs as some studies indicate positive effects on cognitive decline in AD.³² Another interesting result is, that the APOE genotype appears not to be associated with decline. The APOE status and its relation to AD progression is still matter of debate as findings are contradictory.³³

Our study has limitations and strengths. Regarding the former, the medium term observation period has to be mentioned. Alzheimer patients are known to survive 8–10 years after diagnosis. Although 3–4 years of observation is a relevant time span, only further follow-up will allow for mortality related analyses (e.g., PrP^c at diagnosis vs. survival time). Second, declines on the different scales were assumed to be linear and thus estimated by linear regression, as this is a commonly used method. Pre-analyses of the slopes did not contradict this assumption. Still, differently shaped trajectories have been described.³⁴ Sigmoidal or hyperbolic trajectories might also partially explain our findings, that high baseline iADL was associated with faster decline, while worse UPDRSIII or GDS baseline values were associated with slower decline. When analyzing future follow-ups (longer observation periods) this must be considered.

Third, the sample size (n = 84) is rather small. However, this may be compensated by the considerable strength of having used diagnostic criteria supplemented by CSF biomarker analyses. This allows for more accurate diagnoses as compared with the classic, purely clinical 1984 McKhann’s criteria. Therefore, incorrect diagnoses should only make up a very small fraction of the patient group in contrast to the “purely clinical” cohorts in the majority of AD studies.

In conclusion, baseline plasma levels of PrP^c at the time of AD diagnosis appear not to be associated with cognitive or functional deterioration in this study. Thus, plasma PrP^c at baseline seems not to be useful as a biomarker of disease progression. However, Further studies need to be performed in order to verify the validity of these findings.

Materials and Methods

Patients included into this analysis (n = 84) were enrolled in the prospective, observational “rpAD”-study (rpAD, “rapidly progressive Alzheimer’s disease”). Enrollment took place at the time of diagnosis. All patients or legal caregivers (in case patients were cognitively unable to do so) provided informed consent. The study has been approved by the local ethics committee (Georg August University, Göttingen, Germany, Votum No 9/6/08).

Diagnoses were established using research criteria suggested by Dubois et al.,³⁵ meaning that, in addition to the clinical presentation, CSF biomarkers as well as cerebral imaging were studied to achieve more accurate diagnoses. Blood and CSF were acquired as part of the routine diagnostic process. CSF Aβ1-40, Aβ1-42, Aβ ratio (10 × Aβ1-42/Aβ1-40), Tau and pTau were determined by standard methods (Innogenetics, Gent, Belgium), as were APOE and PRNP Codon 129 genotypes. Plasma PrP^c was quantified using modified BetaPrion HUMAN EIA kits by Roboscreen GmbH (Analytik Jena AG), Leipzig, Germany. The proteinase K digestion step was omitted.

All patients underwent a semi-structured interview, neuropsychological testing (CERAD-NP plus Trailmaking Test A³⁶), physical examinations and cerebral imaging (CT or MRI) at the time of diagnosis. Furthermore, global, functional and motor assessment was operationalized using MMSE (extracted from CERAD, 0–30pts.), iADL (0–8pts),³⁷ bADL (0–48 pts),³⁸ GDS (1–7 pts)³⁹ and UPDRSIII (0–108 pts)⁴⁰ scales. Shortly summarized, MMSE measures overall cognitive function in different domains as a screening tool, ADL scales assess the performance in regards to daily activities (i.e., ability to use the telephone, to go shopping, to use the toilet etc.), GDS is a global dementia staging tool and UPDRSIII quantifies extrapyramidal motor dysfunction (i.e., rigidity, bradykinesis, postural instability etc.). These were re-tested annually. Velocity of decline for each of these scales was determined by means of linear regression (least square method in accordance with Villemagne et al.⁴¹) [pts/yr]. Due to the nature of the scales, deterioration/worsening in the particular domains is indicated by a decrease of MMSE, bADL, iADL or an increase in UPDRSIII or GDS.

To uncover potential associations of baseline PrP^c with deterioration in different domains while controlling important possible confounders, multiple regression analyses were applied. Dependent variables were: decline of either MMSE, bADL, iADL, UPDRSIII or GDS. Independent co-variables were: sex, age, education, APOE genotype, PRNP Codon 129 genotype, presence of diabetes mellitus, presence of hypertension, differential pharmacological treatment and baseline values of the dependent variables. Regarding drug intake (memantine, acetylcholineesterase inhibitors or neuroleptics (dopamine antagonists and derivates), the durations of medication use were taken into account by using fractions of the periods of drug intake out of the individual observation periods as co-variables, since these yield more differentiated information than simple binary indicators (drug-intake: yes/no). Therefore, it did not only matter whether the patient took a drug or not, but for how long.

First, a basic model using only the baseline of the dependent, sex, age and education as co-variables was constructed. To build a final model, the remaining co-variables were added to the basic model only in case of statistical significance at a 5% level. Finally the influence of PrP^c was investigated in the resulting model. PrP^c values were log transformed in order to account for the skewness in their distribution. Statistical analyses were performed using the statistical software R (version 2.14.1 for Microsoft Windows, The R Foundation for Statistical Computing).

Abbreviations:
Aβ	=	amyloid beta
AChEI	=	acetyl choline esterase inhibitor
AD	=	Alzheimer disease
ADL	=	activities of daily living
APOE	=	Apolipoprotein E gene
bADL	=	basic ADL
CSF	=	cerebrospinal fluid
est.	=	estimate
f	=	female
GDS	=	global deterioration scale
iADL	=	instrumental ADL
IQR	=	interquartile range
LTP	=	long term potentiation
m	=	male
MMSE, CERAD subscale	=	Minimental Status examination
n.a.	=	not available/not applicable
PRNP C129	=	Prion Protein Gene Codon 129
PrPc	=	physiological form of the Prion protein
pTau	=	hyperphosphorylized Tau
pts	=	points
SD	=	standard deviation
UPDRS III	=	part III of the Unified Parkinsons Disease Rating Scale

Disclosure of Potential Conflicts of Interest

The authors report that no conflicts of interest exist.

Acknowledgments

This work was supported by a Bundesministerium für Bildung und Forschung grant within the German Network for Degenerative Dementia (KNDD-2, 2012–2015, determinants for disease progression in AD, grant No. 01GI1010C).