Cerebrospinal Fluid Amyloid-β Oligomer Levels in Patients with Idiopathic Normal Pressure Hydrocephalus

Study populations

Our studies were performed on two iNPH cohorts: an evaluation cohort (cohort-1) and a validation cohort (cohort-2).

Comparison cohorts

To comparatively evaluate the AβO^10–20 levels in patients with iNPH, we also enrolled neurologically healthy controls (HCs) and patients with other types of neurodegenerative disease (i.e., AD, PD, and progressive supranuclear palsy (PSP)) into comparison cohorts. Sixteen patients with AD were diagnosed by a neurologist according to criteria for probable AD published by the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for AD [11]. Fifteen patients with PD were diagnosed by a neurologist according to the Movement Disorder Society Clinical Diagnostic Criteria for PD [12]. Fourteen patients with PSP were clinically diagnosed by a neurologist according to the Movement Disorder Society criteria [13]. In total, from 14 patients with PSP, we diagnosed 9 of them as Richardson type, 3 as pure akinesia and gait freezing, and 2 as PSP-P. Twenty-seven HCs who were > 60 years of age and had Mini-Mental State Examination (MMSE) scores of 26 points or higher were enrolled. All individuals with AD, PD, PSP, and HCs had previously lumbar puncture for various reasons, such as additional diagnostic information, investigations into headache etiologies, preoperative insertion of a lumbar drainage catheter for aortic surgery, undergoing elective myelography for degenerative lumbar spine disease, or undergoing elective surgery with spinal anaesthesia.

Study design

Our study was designed to include 3 distinctive parts, that we further assigned as Study 1, 2, and 3 (Fig. 1).

OPEN IN VIEWER

Fig. 1

Study designs. First, we compared iNPH cohorts with comparison cohorts: AD, PD, PSP, and HCs (Study 1). Second, we compared iNPH cohort-1’s measurement before and after CSF shunting (Study 2). Third, we subdivided iNPH cohort-1 members into AβO^10–20 decrease or increase subgroups and compared them in terms of biomarkers and neurological statuses at before and 1 and 3 years after CSF shunting (Study 3). AD, Alzheimer’s disease; AβO^10–20, amyloid-β oligomer^10–20; CSF, cerebrospinal fluid; iNPH, idiopathic normal pressure hydrocephalus; HCs, healthy controls; PD, Parkinson’s disease; PSP, progressive supranuclear palsy.

CSF analysis

CSF samples were obtained via lumbar puncture performed in the L3–L4 or L4–L5 interspace with a spinal needle from each individual. In the patients with iNPH, lumbo-peritoneal shunting was performed according to our previously reported method [20]. We used an adjustable valve in all patients (Medtronic Neurosurgery, Goleta, CA, or Integra Codman, Raynham, MA). For postoperative follow-up, we adjusted the valve pressure and monitored proper shunt function with CSF withdrawal via valve reservoir puncture as previously reported and postoperative CSF samples in the patients with iNPH were obtained in the same way [8, 20]. If shunt malfunction was detected, the patient underwent shunt revision immediately.

To remove cells and debris, CSF samples were centrifuged at 1,690 g for 10 min at 4°C and then aliquoted. Hemorrhagic samples which had an erythrocyte count above 500/μL were excluded. The samples were divided into 1 ml polypropylene tubes and stored at –80°C till the assay. We thawed samples before the assay and the remaining in the tube was thrown away after each assay, therefore we had only one freeze-thaw cycle. The storage periods ranged from CSF collection to the assay time and varied among samples. CSF samples of iNPH cohort-1 was collected between April 2011 and December 2020 (median days, 1st and 3rd quartile that the samples stored between the initial evaluation to the assay was 1,245 [989.75 –1,848]), iNPH cohort-2 –between December 2016 and March 2019 (634 [387 –775]), AD group - between October 2008 and December 2014 (1,946.5 [1,775 –2,306.25]), PD group - between April 2010 and July 2014 (2,276 [2,080 –2,930.5]), PSP group - between May 2007 and July 2015 (2,695.5 [2,130.5 –4,025.75]), and HCs group - between October 2006 and June 2016 (2,695 [2,062 –3,928]). We used ELISA kits to measure the levels of tau (T1006; Nipro Corporation, Osaka, Japan), pTau (T1008; Nipro Corporation, Osaka, Japan), Aβ₃₈ (code no.: 27717; IBL Japan, Fujioka, Japan), Aβ₄₂ (INNOTEST; code no.: 81583; Fujirebio, Tokyo, Japan), the toxic Aβ_42conformer (code no.: 27709; IBL Japan), and AβO^10–20 (code no.: 298-80101; FUJIFILM Wako Pure Chemical Corporation, Osaka Japan). For the AβO^10–20 ELISA kit, BAN50 was used as both the capture antibody and detector antibody. This kit specifically detects Aβ oligomers containing 10–20 monomers (40–200 kDa) and does not detect monomers, dimers, trimers, tetramers, or hexamers [7]. Luminescence measurements were used to detect our target AβO^10–20 given that these represent < 1%of AβO^10–20 overall. Therefore, we used a 96-well microplate reader (SpectraMax L; Molecular Devices Japan, Tokyo, Japan) and its accompanying software (SoftMax Pro 5.4.8). The lower limits of detection of ELISA kits for AβO^10–20, Aβ₄₂, Aβ₃₈, Aβ₄₂ toxic conformer, pTau, and tau were 0.41 pM, 62.5 pg/mL, 9.38 pg/mL, 3.13 pg/mL, 25 pg/mL, and 75 pg/mL respectively. We used two wells for each sample in the assay. All kits we used were of the same lot number. Coefficient variations were lower than 10%.

Statistics

Data distribution was evaluated graphically using histograms and Q-Q plots. In study 1, one-way ANOVA followed by the Tukey honestly significant difference test was performed in age and biomarkers. Kruskal-Wallis test followed by the Steel-Dwass test was performed in MMSE scores. Fisher’s exact test was conducted to compare proportions. Areas under receiver operating characteristic curves (AUROCs) were calculated as goodness-of-fit descriptors for predictors of iNPH status versus AD or PD and PSP status. In study 2, the Friedman’s test followed by Wilcoxon signed-rank test with Holm’s correction was conducted to evaluate differences between neurological statuses before and 1 and 3 years after CSF shunting, and the Paired T test was conducted to evaluate differences between biomarkers measured before and after CSF shunting. In study 3, T-test was performed in biomarkers levels and Evans Index and Mann-Whitney’s U-test was performed in MMSE scores, FAB scores, mRS scores, and scores in iNPHGS. Fisher’s exact test was performed in sex ratios, DESH prevalence, the percentages of patients with DSWMH scores of 0–1, comorbidity prevalence and percentages of patients improved in MMSE. All statistical analyses were performed with statistical software EZR version 1.41 based on R and R commander [21]. Results are shown as means±standard deviations for ages and CSF biomarker values and as medians and interquartile ranges for clinical evaluation battery scores. A p-value < 0.05 was considered statistically significant.

Ethics

This study was approved by the Ethics Committee of Juntendo University (approval no.: 17-022). All study participants or their legal guardians were briefed on the details of the study and provided written informed consent.

Comparisons of the iNPH cohorts with the HCs, AD, PD, and PSP groups (Study 1)

Comparisons of the iNPH cohorts and the HCs, AD, PD, and PSP groups revealed significant differences in the age, the sex ratios, and the proportions of patients with hypertension, diabetes, and dyslipidemia. From the total of 8 patients with previous stroke history, 5 patients of them had symptomatic lacunar infarction, 2 patients had symptomatic cerebellar infarction, and 1 patient had asymptomatic lacunar infarction, and they recovered neurologically without sequelae that can affect diagnosis and neurological outcome. The AβO^10–20 levels in the iNPH cohort-1 and the AD group were significantly higher than in the HCs, PD, and PSP groups (Fig. 2). The Aβ₄₂ levels in the AD group were significantly lower than those in iNPH cohort-1 and the HCs (p = 0.014, < 0.001 respectively). The pTau levels in the AD group were significantly higher than those in any other group (Table 1). An AβO^10–20 cut-off level of 6.392 pM distinguished iNPH cohort-1 members from AD group members with 68.8%sensitivity, 65.6%specificity, and an AUROC of 0.678. An AβO^10–20 cut-off level of 5.789 pM distinguished iNPH cohort-1 members from PD and PSP group members with 96.6%sensitivity, 81.2%specificity, and an AUROC of 0.944 (Fig. 3).

OPEN IN VIEWER

Fig. 2

AβO^10–20 levels in each group. AβO^10–20 levels for each individual group are represented in a scatter plot. The vertical axis shows the CSF concentrations of AβO^10–20. Significance testing was done with the One-way ANOVA followed by the Tukey honestly significant difference test and is indicated as ^**p < 0.01 or ^***p < 0.001 (Versus iNPH cohort-1) and ^# #p < 0.01 or ^# # #p < 0.001 (Versus iNPH cohort-2). AD, Alzheimer’s disease; AβO^10–20, amyloid-β oligomer^10–20; iNPH, idiopathic normal pressure hydrocephalus; HCs, healthy controls; PD, Parkinson’s disease; PSP, progressive supranuclear palsy.

OPEN IN VIEWER

Fig. 3

AUROC analysis in study 1. AUROC analysis of AβO^10–20 levels (blue), Aβ₄₂ levels (green), and pTau levels (purple) as tools for differentiating groups. A) AβO^10–20 levels differentiated members of iNPH cohort-1 from members of the AD groups with an AUC value of 0.678. B) AβO^10–20 levels differentiated members of the iNPH cohort-1 from members of the PD and PSP groups with an AUC value of 0.944. AD, Alzheimer’s disease; AUC, area under the curve; Aβ42, amyloid beta 42; AβO^10–20, amyloid-β oligomer^10–20; CI, confidence interval; iNPH, idiopathic normal pressure hydrocephalus; PD, Parkinson’s disease; PSP, progressive supranuclear palsy; pTau, phosphorylated tau.

In the validation analysis, the iNPH cohort-2 displayed significantly higher levels of AβO^10–20 than the HCs, PD, and PSP subgroups did, while did not display any significant difference against AD. The result was consistent with the observed for iNPH cohort-1.

Comparisons of measurements recorded before and after CSF shunting in iNPH cohort-1 (Study 2)

In the iNPH cohort-1, gait disturbance was significantly improved and maintained for 3 years after CSF shunting. Cognitive function was significantly improved at 1 year after CSF shunting; however, these improvements were not maintained for 3 years.

In terms of biomarkers, Aβ₃₈, Aβ₄₂, and Aβ₄₂ toxic conformer levels were significantly increased and AβO^10–20 levels were significantly decreased relative to their pre–CSF shunting levels.

Comparison between AβO^10–20 decrease and AβO^10–20 increase subgroups (Study 3)

At baseline, AβO^10–20 increase subgroup had lower levels of FAB score, Aβ₄₂ toxic conformer, and AβO^10–20. There was no significant difference in age, proportion of male, mRS scores, iNPHGS scores, Evans index, proportion of the patients with DESH, DSWMH score of 0 or 1, and comorbidities such as hypertension, diabetes, dyslipidemia, cardiac diseases, and stroke at the baseline (Supplementary Table 2). FAB scores in AβO^10–20 increase subgroup were still lower than AβO^10–20 decrease subgroup at 1 year after CSF shunting but the significance diminished at 3 years after CSF shunting. There was no significant difference in mRS score and iNPHGS scores at 1 and 3 years after CSF shunting. Although there was no significant difference at MMSE scores at 3 years after CSF shunting, proportion of patients who improved or maintained in MMSE scores were significantly higher in AβO^10–20 decrease subgroup (Table 3, Fig. 4).

OPEN IN VIEWER

Fig. 4

Cognitive outcomes in AβO^10–20 decrease and increase subgroups. A distribution of preoperative and postoperative grade of dementia severity according to MMSE scores in AβO^10–20 decrease and increase subgroup displayed as the table. Numbers in each box indicate the number of patients. Blue boxes indicate patients in “good outcome” and orange boxes indicate patients in “poor outcome”. AβO^10–20, amyloid beta oligomer^10–20; CSF, cerebrospinal fluid; G, grade.

An AβO^10–20 cut-off level of 5.648 pM distinguished AβO^10–20 decrease subgroup members from AβO^10–20 increase subgroup members with 55.6%sensitivity, 95.7%specificity, and an AUROC of 0.691 (Supplementary Figure 1).

AβO^10–20 levels are elevated in patients with iNPH

In iNPH, CSF stagnation is one of the causes of Aβ deposition followed by cognitive impairment [9]. According to a previous report, CSF concentrations of Aβ₄₂ in patients with iNPH are lower than or equal to those observed in HCs [22]. Hypometabolism of AβPP, dilution due to CSF accumulation, and aggregation of Aβ are considered the causes of these changes [23, 24]. Similarly, accelerated production and deposition of Aβ is considered as one of the hallmarks of AD pathology [25, 26]. Aβ₄₂ has a tendency to aggregate and forms senile plaques and protofibrils in the brain of patients with AD [7]. However, cognitive impairment severity correlates poorly with Aβ plaque loads, but strongly with the levels of soluble Aβ oligomers [27].

The AβO^10–20 levels in the iNPH cohort-1 were significantly higher than those in the HCs and the patients with PD or PSP. This result was validated with the data from iNPH cohort-2. Consistent with the findings of a previous research, the AD group had higher AβO^10–20 levels than HCs [7]. However, there was significant difference in age among groups. Although it could affect biomarker profile in this analysis, there was no significant correlation between age and biomarker levels in each group (Supplementary Table 1) Contrary to our expectations, AβO^10–20 levels in the patients with iNPH were approximately the same as in the patients with AD. This result can be interpreted in the sense that Aβ aggregation may be up to the same oligomer chain length in both the iNPH group as in the AD group. Although the antibody we used in the present study detects mainly 10–20-monomer oligomers [7], we were unable to evaluate their molecular weight. Therefore, we cannot exclude the possibility that larger Aβ chain aggregation produces larger macromolecules and apparently decreases the total oligomer numbers in a standard volume of the CSF sample.

Patients with iNPH sometimes show clinical symptoms that are similar to those of other neurodegenerative diseases, which complicates differential diagnosis. According to past reports, AD-related pathology can be observed in 18%–75%of patients with iNPH [28, 29], and 11%–86%of patients with iNPH exhibit parkinsonism [30–33]. Our results indicate that AβO^10–20 are not good diagnostic biomarkers for differentiating iNPH from AD but are useful for differentiating iNPH from PD and PSP. However, as has been previously reported, AD can be differentiated from iNPH using CSF levels of pTau.

AβO^10–20 decrease after CSF shunt placement contribute to cognitive improvement in iNPH patients

It has been established that CSF production is affected in iNPH, and CSF concentrations of AβPP-derived peptides increase after shunt surgery [34, 35]. It is possible that CSF elimination following shunt surgery improves CSF production and AβPP metabolism in the periventricular space [36–38]. In the present study, Aβ₃₈ and Aβ₄₂ levels in the iNPH cohort-1 members increased after CSF shunting as previously reported, and that may reflect an improvement in CSF turnover. As we expected, AβO^10–20 levels decreased after CSF shunting. These results support our speculation that AβO^10–20 accumulation may be caused by CSF stagnation, and it is probably relieved by CSF shunting. Aβ aggregation as an iNPH pathophysiological phenomenon may be restored to normal by the improved clearance of Aβ peptides.

As in AD, cognitive impairment in iNPH may also be affected by AβO^10–20 levels. In the present study, AβO^10–20 decrease subgroup showed better cognitive outcome at 3 years after shunt surgery than AβO^10–20 increase subgroup and higher levels of AβO^10–20 can predict AβO^10–20 decrease with AUC of 0.691. AβO^10–20 increase subgroup members had lower levels of FAB scores, AβO^10–20 and Aβ₄₂ toxic conformer at the baseline. Although there was no significance, AβO^10–20 increase subgroup members had lower levels of MMSE score and higher proportion of patients with DSWMH score 2–4. These findings imply other concomitant pathology and probably vascular burden might have affected the results in the AβO^10–20 increase subgroup members.

In the present study, iNPH cohort-1 members showed higher levels of Aβ₄₂ toxic conformer than that we had previously reported. We considered that higher levels of Aβ₄₂ in iNPH cohort-1 members leaded to such elevation because the levels of Aβ₄₂ toxic conformer ratio were almost same as those we previously reported [8].

Study limitations

The main limitations of the current study were that we had only two cohorts of patients with iNPH and that the number of recruited participants was low. During patient recruitment, we excluded a number of patients from iNPH cohort-1 for several reasons, and these exclusions might have potentially affected the results of the present study. The exclusion of patients who had less than 3 years follow-up period can potentially be a reason to overestimate the outcome in iNPH cohort-1 because neurologically deteriorated patients might have been unable to visit our hospital. The exclusion of patients who had > 50 pg/ml of CSF pTau level also can be a reason to overestimate the outcome in iNPH cohort-1; however, their exclusion intended to remove the undesirable effect of comorbid AD pathophysiology.

Some parameters at baseline differed among enrolled cohorts in study 1, such as age and proportion of male patients and those with hypertension, diabetes, and dyslipidemia. Although there was no significant correlation between age and biomarker levels, these differences could have affected neurological outcome and biomarker profile. As aforementioned, although we excluded patients with iNPH who had past ischemic events, and there were no significant differences between the AβO^10–20 decrease subgroup and AβO^10–20 increase subgroup in the proportions of patients with DSWMH scores of 2–4, we did not do further investigations for vascular pathologies, such as performing cerebral blood flow studies. Therefore, vascular lesions may have affected biomarker measurements and neurological assessments.

Although we considered that amelioration of CSF stagnation may improve AβO^10–20 accumulation, we could not quantify CSF flow or the volume of CSF eliminated through the shunt device. Future research should attempt to evaluate the amount of CSF produced and eliminated to test this consideration.