|Home | About | Journals | Submit | Contact Us | Français|
Definitive diagnosis of Alzheimer's disease (AD) can only be made by histopathological examination of brain tissue, prompting the search for premortem disease biomarkers. We sought to determine if the novel brain injury biomarker, Visinin-like protein 1 (VLP-1), is altered in the CSF of AD patients compared to controls, and to compare its values to the other well-studied CSF biomarkers 42-amino acid amyloid-β peptide (Aβ 1-42), total Tau (tTau), and hyperphosphorylated Tau (pTau).
CSF samples from 33 AD patients and 24 controls were analyzed by ELISA to measure concentrations of Aβ1-42, tTau, pTau, and VLP-1. The diagnostic performance of these biomarkers was compared using receiver operating characteristic (ROC) curves.
CSF VLP-1 concentrations were significantly higher in AD patients (365 ± 166 ng/L, median ± inter-quartile range) compared to controls (244 ± 142.5). While the diagnostic performance of VLP-1 alone was comparable to that of Aβ, tTau, or pTau alone, the combination of the 4 biomarkers demonstrated better performance than each individually. VLP-1 concentrations were higher in AD subjects with APOEε4/ε4 genotype (599 ± 240 ng/L) compared to ε3/ε4 (376 ± 127) and ε3/ε3 genotypes (280 ± 115.5). Furthermore, VLP-1 values demonstrated a high degree of correlation with pTau (r = 0.809) and tTau (r = 0.635) but not Aβ1-42 (r = -0.233). VLP-1 was the only biomarker that correlated with MMSE score (r=-0.384, p=0.030).
These results suggest that neuronal injury markers like VLP-1 may have utility as biomarkers for AD.
The diagnosis of Alzheimer's disease (AD), the most common form of dementia in Western countries, is largely based on historical and clinical criteria. Although many studies report a reasonably high degree of diagnostic accuracy (80-90%), often these studies include patients evaluated at specialized centers with advanced disease (1). At present, post-mortem examination of brain tissue is the only tool for definitive diagnosis. Therefore, the development of a biomarker for AD would aid greatly in the diagnosis of this disease. In addition, such a marker could potentially be utilized to measure efficacy in future therapeutic trials.
Most studies of AD biomarkers have focused on known pathological substrates for the disease. Amyloid plaques and neurofibrillary tangles are pathological hallmarks of AD (2), and are comprised primarily of abnormally aggregated endogenous proteins. Amyloid plaques (extracellular proteinaceous aggregates) are principally composed of the amyloid-β peptide (Aβ), a 38-42 amino acid peptide fragment of the amyloid precursor protein (APP). The major species, the 42-amino acid peptide (Aβ1-42) (3, 4), is significantly decreased in the CSF of patients with AD (5-8). Neurofibrillary tangles are intraneuronal protein aggregates found mainly in neurites and primarily composed of hyperphosphorylated Tau (pTau), a microtubule-associated protein. Several studies have shown that total Tau (tTau) and pTau are increased in CSF from AD patients (9-12). Still, substantial overlap in values for these biomarkers between cases and controls limits their utility as diagnostic biomarkers.
Another class of biomarkers that may have utility in the diagnosis of AD are biomarkers that reflect neuronal death rather than specific markers of disease pathogenesis. Such markers may provide information about disease progression related to functional outcome, and may have utility in future clinical trials testing therapeutic efficacy. Several reports have demonstrated the lack of correlation between amyloid plaque load and the degree of dementia, suggesting that the former may not directly relate to the latter (13, 14). Therefore, a neuronal death biomarker might have greater correlation with dementia severity compared to the well-studied pathological biomarkers.
We have recently identified several potential biomarkers for brain injury and have characterized one of these markers in acute ischemic stroke patients (15). This biomarker, Visinin-like protein 1 (VLP-1), is a calcium sensor protein which is expressed in high abundance in neurons of the central nervous system (16, 17). VLP-1 is increased in the CSF of rats following transient focal ischemia, and is detectable in increased concentrations in the plasma of ischemic stroke patients (15). In this study, we examined the possibility that this novel biomarker of brain injury might be altered in AD.
All patients underwent a thorough clinical evaluation, which included a medical and family history, physical, neurologic, psychiatric, and minimental status examinations (MMSE), performed by a dementia specialist (NA). Electrocardiogram, electroencephalopgram, and head CT was also performed. 33 patients with a clinical diagnosis of AD (NINCDS-ADRDA criteria for probable AD) (18) were included in this study. 24 healthy controls also participated and were free of neurological or psychiatric disorders. APOE genotyping was performed by mini-sequencing, as described previously in detail (19). All patients (or their nearest relatives) and controls gave informed consent to participate in the study, which was conducted in accordance with the provisions of the Helsinki Declaration. The Ethics Committees at Karolinska Institute and Göteborg University approved the study.
The mean age and gender distribution of subjects in each group did not differ significantly (Table 1). The mean duration of disease in the AD group was 3.9 years. As expected, the MMSE scores were significantly lower in the AD group (23.0 vs. 29.8, p < 0.001).
CSF samples were collected into polypropylene tubes by lumbar puncture at the L3/L4 interspace. Immediately after collection, a cell count was performed by light microscopy in Bürker chambers, and all samples had less than 500 erythrocytes/μl. The remaining CSF was centrifuged at 2000g for 10 min (to eliminate cells) and frozen in aliquots at -80°C
CSF samples were assayed for total Tau (tTau), hyperphosphorylated Tau (pTau at Thr-181), and Aβ1-42 using sandwich ELISAs as previously described (11, 20, 21). CSF VLP-1 was measured using a sandwich ELISA (monoclonal antibody for capture and rabbit polyclonal antibody for detection) as previously described (15).
Differences in patient characteristics and biomarkers between two groups were compared using chi-square test or student's t-test as appropriate. ANOVA with a post-hoc Tukey's test was used for comparisons between multiple groups. The correlations between CSF VLP-1 and other markers were also assessed with partial correlation coefficients, which measure the strength of a relationship between two variables, while controlling for the effect of AD status. The diagnostic ability of these biomarkers was evaluated using receiver operating characteristic (ROC) curves which plot true positive rates (sensitivity) versus the false positive rates (1-specificity) across all possible thresholds. As a global measure for the accuracy of diagnosis, the area under ROC curve (AUC) was also calculated for each individual biomarker (22). All statistical comparisons were performed using the statistical package SAS (version 9) while all ROC analyses were performed with ROCKIT, a widely used freeware available from the Kurt Rossman Laboratories at the University of Chicago. A p-value less than 0.05 was considered significant and all statistical tests were two-sided.
CSF tTau and pTau values were significantly higher in AD patients compared to controls (p < 0.001 for both, Table 1). In addition, Aβ1-42 values were lower in AD patients compared to controls (p < 0.001, Table 1), consistent with numerous other studies (9-12). VLP-1 concentrations in the CSF were significantly higher in the AD patients compared to controls (365 ± 166 vs. 244 ± 142.5, p < 0.001, Fig. 1).
Despite the significant difference between VLP-1 values in the CSF of controls vs. AD patients, considerable overlap was observed (similar to Aβ and Tau). To see if VLP-1 provides additional utility to the diagnosis of AD above and beyond the contribution of Aβ, tTau, or pTau alone, we performed a receiver operating characteristic (ROC) analysis for each individual biomarker alone compared to the combination of all biomarkers. The area under the ROC curves for VLP-1, Aβ, tTau, pTau, and an optimum linear combination of all biomarkers are shown in Fig. 2. The area under the curve (AUC) was similar between all biomarkers individually; however, the linear combination of all biomarkers resulted in an approximately 5% improvement (Fig. 2).
To examine possible relationships between CSF VLP-1 values and patient characteristics, we performed correlation analyses between VLP-1 and age, disease, duration, MMSE, and the number of APOE ε4 alleles. VLP-1 correlated with MMSE and the number of APOE e4 alleles (Fig 3A). None of the other biomarkers correlated with MMSE in this patient population (Aβ1-42, r = 0.350, p = 0.497; tTau, r = -0.295, p = 0.100; pTau, r = -0.202, p = 0.264). To further examine the relationship between APOE genotype and CSF VLP-1 concentrations, we calculated mean CSF VLP-1 values by different genotypes. APOE ε4/ε4 individuals had the highest concentrations, followed by ε3/ε4 and ε3/ε3 individuals (Fig. 3B).
To examine if VLP-1 concentrations in the CSF were related to values of the other biomarkers studied, we performed correlations between VLP-1 and tTau, pTau, or Aβ1-42 using data from both AD patients and controls (Fig 4). VLP-1 and pTau showed the greatest correlation (Fig 4C, r = 0.809), while Aβ1-42 did not correlate with VLP-1 (Fig 4A, r = -0.233). Individual correlations for AD patients analyzed separately from controls were also performed, and revealed results similar to that of the total patient population: VLP-1 vs. Aβ1-42 was not statistically significant (r = -0.29671 and -0.1698 in AD and controls respectively), while VLP-1 vs. tTau (r = 0.6221 and 0.7247 in AD and controls) and pTau (r = 0.8747 and 0.6227 in AD and controls) were significantly correlated in the AD and control populations analyzed separately.
In this study we demonstrated that CSF values of VLP-1 are significantly higher in AD patients compared to control subjects. However, like other well-studied biomarkers for AD (Aβ or Tau), there is substantial overlap in values between cases and controls. By ROC analysis (area under the ROC curve) the individual biomarkers alone (Aβ, tTau, pTau, and VLP-1) were roughly equivalent with regard to their diagnostic accuracy. However, when combined together, the diagnostic accuracy increased, suggesting added benefit of multiple biomarkers. Clearly, VLP-1 is not a biomarker that is specific for AD. Indeed, its utility was pioneered in brain injury caused by acute ischemic stroke (15). It is likely that measures of VLP-1 reflect neuronal injury with subsequent release of this intracellular protein into the CSF. Thus, there may be utility in combining biomarkers that reflect different aspects of disease pathogenesis. Both Aβ and Tau reflect different pathological features of AD, while VLP-1 may reflect the end result of the disease—neuronal cell death.
VLP-1 is a cytoplasmic calcium-sensor protein found almost exclusively in neurons of the CNS, and is widely expressed throughout the brain (16, 17, 23), but undetectable in other peripheral tissues (15, 24). We have previously found that VLP-1 concentrations increased in the CSF of rats subjected to middle cerebral artery occlusion, but were undetectable in sham-operated controls. Moreover, VLP-1 was detected in the serum of acute ischemic stroke patients but not in normal blood donors (15). The presumed mechanism of appearance is via leakage from injured or dying neurons into the CSF and peripheral blood. Thus, in acute brain injury high concentrations (up to 1000 μg/L) are detectable in plasma samples (15). However, in neurodegenerative disorders, VLP-1 concentrations are likely to reflect the equilibrium between the protein released from dying neurons and clearance of the protein from CSF. The lower values detected in the CSF of AD patients support this contention.
If VLP-1 is a biomarker of neuronal loss, one might expect that its concentration may correlate with dementia severity. Indeed, in this cohort of AD patients, VLP-1 was the only biomarker that did correlate with MMSE (albeit weak). Others have reported weak correlations between Aβ1-42 or Tau and a variety of dementia severity scores (25-27). It is likely that we were unable to find similar correlations because of our relatively small sample size. Future studies in larger and more well-characterized populations using more sensitive dementia scores will be needed to better determine any relationship between neuronal injury biomarkers and dementia.
A variety of other brain-injury biomarkers have been examined in the CSF of patients with dementia, including neuron-specific enolase (28, 29), S100B (30), and GFAP (31), all with variable diagnostic specificity and sensitivity. More recently, proteomic profiling has also resulted in the identification of several candidate biomarkers (32), including heart-fatty acid binding protein (33, 34), Park 7, and nucleoside diphosphate kinase A (35). The effectiveness of a fluid biomarker is dependent on a multitude of factors, including organ specificity, accumulation in accessible body fluids, stability, clearance, and detectability. Direct comparisons between biomarker candidates will be important to identify such an ideal biomarker. Although in the current study, we did not perform direct comparisons to other candidate biomarkers, it will be important to do so in the future.
APOE genotype is the strongest known genetic risk factor for the development of late-onset AD, with the ε4 allele incurring greatest risk (36-38). The molecular mechanism for this risk is not known; however, it is believed that ApoE protein may play a role in Aβ transport/clearance (39), and that the genotype may also impart increased vulnerability to a variety of CNS injuries (40). Consistent with the latter contention, we have found that AD patients with APOE ε4/ε4 genotypes had the highest concentrations of CSF VLP-1 compared to ε3/ε4 and ε3/ε3 genotypes. Indeed ε4/ε4 individuals had more than twice the concentration of CSF VLP-1 compared to ε3/ε3 individuals. At face value, these results suggest that the ε4 genotype increases vulnerability to neuronal death; however, it is also possible that APOE genotype influences plaque load, which may also influence neurodegeneration.
Our finding that CSF VLP-1 values correlated highly with pTau but not Aβ values in our patient population is very interesting in light of the relationship between Aβ and Tau. Dementia severity appears to correlate with the number of neurofibrillary tangles, but not to the degree of plaque deposition (13). The close correlation between VLP-1 and pTau concentrations in the CSF of AD patients is consistent with these findings, as is the lack of correlation with Aβ.
There are several limitations to this study. First, the number of patients in both control and disease groups is limited. Further studies will be needed to confirm our findings in larger, more well-characterized populations. Second, because the diagnosis of AD was made by clinical criteria, there will undoubtedly be a small but significant group of patients that were misdiagnosed (10-20%), (1). This may account for some of the overlap in values for CSF biomarkers. ApoE genotyping in the control group might help with this diagnostic uncertainty. A much more rigorous study would require autopsy confirmation of diagnosis. Third, our study is limited to a comparison of VLP-1 levels in AD patients vs. controls, a situation that is unlikely to occur clinically. A more relevant comparison should be made across patients carrying the differential diagnosis of dementia. Finally, our CSF samples represent a single “snap-shot” in AD pathogenesis; further studies will be required to understand the time-course or biomarker evolution with disease pathogenesis.
We thank Nancy Brada for recombinant protein production and Mary Jane Eichenseer for help with immunoassay development.
GRANT/FUNDING SUPPORT Supported in part by a grant from Siemens Healthcare Diagnostics (J.H.L.), NIH NINDS R01 NS48283 (J.-M.L.), the American Health Assistance Foundation (J.-M.L.), the Sahlgrenska Hospital, and the Swedish Research Council (project #14002, K.B and N.A.).
FINANCIAL DISCLOSURES O.L., V.M. and J.H.L. are named as co-inventors on pending patents filed by Washington University concerning brain biomarkers. J.H.L. is a consultant to Siemens Healthcare Diagnostics.