Design

A retrospective, descriptive design was used to analyze extracted data from 54 participants. The extracted data were from two databases with information related to AD, one for electrophysiological data and the other for clinical data. The databases resulted from a collaborative agreement between two research laboratories in a northeast metropolitan city. Participants provided electrophysiological data after providing the clinical data. The present study, a secondary analysis, was approved and received exemption status from the institutional review boards (IRBs) for New York University and New York University School of Medicine. This secondary analysis has ethical approval for informed consent exemption from New York University and New York University School of Medicine.

The extracted baseline (BL) and follow-up (FU) data were analyzed collectively. The FU data had a minimum of two years post-BL. A two-year time interval between BL and FU was selected to ensure enough time for the conversion from normal functioning to diagnoses within the AD continuum, facilitating the identification and analysis of the electrophysiological biomarkers. The median time from BL to FU was 4 years (range = 2–14 years) for the extracted data. There was minimal variability in the data collection over the follow-up period. The procedures/protocols, equipment, and assessment tool were consistently maintained. The software quantifying the rsEEG data was assessed annually to ensure accuracy and consistency. The EEG technicians and clinicians were the same with the same director of the Brain Research Laboratories overseeing data collection over the follow-up period. Therefore, the data obtained from participants over the years was methodologically consistent.

Participants

The sample consisted of 54 older adults who were either normal functioning or diagnosed with prodromal AD or an early stage of AD at BL. The sample had a mean age of 67.3 ± 8.1 (range = 53–85). A diagnosis of dementia of the Alzheimer type (DAT) was based on the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM IV) [36] and probable AD according to the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) criteria [37]. The sample was grouped according to the stages of the Global Deterioration Scale (GDS). Exclusion criteria included history of head trauma, substance abuse, signs of abnormal structural brain pathology, diagnosis of delirium, depression, or vascular dementia, and failure to discontinue any psychotropic or other cognitive-acting medication at least 2 weeks prior to the evaluation period.

Measurements

The GDS [38] categorized the sample by overall cognitive decline from normal functioning to a diagnosis of AD. The GDS is a reliability and validity scale to measure the decline of AD [39]. The GDS consists of seven stages. Stage GDS 1, normal functioning, relates to the absence of any subjective complaints or evidence of objective complaints about cognitive impairment. Stage GDS 2, preclinical or early MCI, relates to subjective complaints without evidence of objective cognitive impairment. Stage GDS 3, prodromal or MCI, relates to mild complaints of objective cognitive impairment. Stages GDS 4–GDS 7 meet the DSM IV criteria for DAT. They range from mild AD to severe AD with increasing severity.

Brain activity changes were measured with the EEG. The raw rsEEG data were quantified with Neurometrics [40]. The electrophysiological biomarkers were characterized as rsEEG delta (1.5–3.5 Hz), theta (3.5–7.5 Hz), alpha (7.5–12.5 Hz), and beta (12.5–25 Hz) frequencies [6, 7]. Neurometrics is a reliable method for quantifying the raw rsEEG data from normal-functioning older adults and those with AD [40].

A population-based normative database was used to determine the changes in the electrophysiological biomarkers (rsEEG delta, rsEEG theta, rsEEG alpha, and rsEEG beta frequencies). Using a normative database obviates the need for a control group since the database consists of the biomarkers of individuals without any pathology and from the general population. This normative database is reliable, replicable, and ethnic-free for normal-functioning older adults to those living with a diagnosis within the AD continuum [41]. The normative database is also reliable when determining the changes in the electrophysiological biomarkers for schizophrenia [41], obsessive-compulsive disorder [41], and depression [23].

Procedures

Electrophysiological data collection

Participants were seated in a light-attenuated, soundproof room. Nineteen standard Ag/AgCl recording electrodes were attached to the scalp using water-soluble paste, according to the 10/20 International System: Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T3, T4, T5, T6, Fz, Cz, and Pz (see Figure 1) [40].

Display full size

Figure 1. 

The 10/20 International System for electrode placement in electroencephalogram. The electrode placement indicates the cortical area for the recording of the brain activity. Frontal cortical area = Fp1, Fp2, F3, F4, F7, F8, Fz. Temporal cortical area = T3, T4, T5, T6. Parietal cortical area = C3, C4, Cz, P3, P4, Pz. Occipital cortical area = O1, O2. Reference electrode: A1 = left ear; A2 = right ear

Electro-oculogram electrodes (differential eye channels) were diagonally placed above and below the eyes’ orbit [40] to detect eye movement artifacts. An additional electrode was placed on the thorax to detect cardiac rhythm (EKG). A grounding electrode was placed on the dominant arm. Recordings were collected and referenced to linked earlobes. The EEG amplifiers had a bandpass from 0.5 to 30 Hz (3 dB points) with a 60 Hz notch-filter. Impedances were maintained below 5,000 ohms. The data was digitized with the A/D converter at 200 Hz with 12-bit resolution. Data acquisition was performed on a Spectrum 32 EEG Acquisition System (Cadwell Laboratories, Kennewick, WA.)

Trained EEG technicians collected 20 minutes of eye-closed raw EEG data while the participant remained at rest. This is also known as rsEEG data. The EEG technicians monitored the participant and the EEG recording for artifacts such as sleepiness, muscle movement, or eye movement. The EEG technicians visually edited EEG data for the artifacts because they may affect the quantitative EEG (qEEG) analysis. This quality check procedure was supplemented with an automatic EEG artifact detection algorithm. From the 20 minutes of collected rsEEG data, only 1–2 minutes (24–48 artifact-free epochs, 2.5 seconds long) were randomly selected for qEEG analysis.

EEG data analysis

All 19 monopolar derivations, absolute (ABS) power, relative (REL) power, mean frequency (MMF), inter- and intra-hemispheric coherence, bilateral symmetry, and intra-hemispheric gradients of power (qEEG dimensions) were computed and characterized as rsEEG delta (1.5–3.5 Hz), rsEEG theta (3.5–7.5 Hz), rsEEG alpha (7.5–12.5 Hz), rsEEG beta (12.5–25 Hz), and total power. Neurometrics is previously described in detail [40]. The essential features are briefly provided. Power spectral analysis was performed on the artifact-free raw rsEEG data using Fast Fourier Transform (FFT) and log transformation to achieve Gaussianity [40]. Univariate analysis was used to statistically evaluate the relative distributions of every biomarker [40]. Age regressions were used to control for any variance from age [40], and Z-transformation was used to standardize the values relative to age-expected normal values. The Z-transformation includes this step: $\text{Mean Ss – }\frac{\text{Pop Age Norm}}{\text{Pop Age SD}}$, where Ss means the sum of squares, Pop means population, Norm means expected or norm, SD means standard deviation [40]. The extracted biomarkers are age-matched and statistically analyzed to the population-based normative database.

Statistical analyses

Data were analyzed with SPSS 28.0 (Windows). The study had a power of 0.80 to detect a moderate effect size at an a priori alpha of 0.05 (adjusted for post-hoc analyses). Parametric and nonparametric statistics were used to analyze the data distribution, demographics, and electrophysiological biomarkers, all at BL and FU. Descriptive statistics were used to characterize age and gender. Kruskal-Wallis test was used to analyze the difference among the GDS stages, age, and gender at BL and FU. Group average topographic brain images were only used to display the distribution of electrophysiological biomarkers.

BL evaluations

The sample consisted of 23 males (43%) and 31 females (57%) with an average age of 67.3 ± 8.1 (range = 53–85) (see Table 1). At BL, 21 (39%) participants were normal functioning (GDS 1 and GDS 2), 20 (37%) diagnosed with MCI (GDS 3), and 13 (24%) diagnosed with probable AD (GDS 4 and GDS 5). Age and gender were not significantly different across GDS stages (age: χ² = 0.94, df = 4, P ≥ 0.7; gender: χ² = 1.5, df = 4, P ≥ 0.83).

The brain images indicate that only ABS rsEEG theta and REL rsEEG theta electrophysiological biomarkers were increasing across the GDS stages from the frontal to the posterior region (P ≤ 0.01) of the head (see Figure 2). The MMF rsEEG delta significantly increased from GDS 4 and 5 (P ≤ 0.01) while the ABS total power and REL rsEEG alpha remained within normal ranges. The absence of a significant change in REL rsEEG theta from REL rsEEG alpha suggests that the increasing rsEEG theta is not a result of a decrease in rsEEG alpha-it is an independent process.

Display full size

Figure 2. 

Topographic brain images in Global Deterioration Scale stages 1–5 at baseline (BL). MMF: mean frequency; ABS: absolute; REL: relative; GDS: Global Deterioration Scale

Cortical areas with electrophysiological changes correspond to the electrode placements as diagrammed in Figure 1. MMF delta = mean frequency delta, ABS theta = absolute theta, REL theta = relative theta, REL alpha = relative alpha, and ABS total = absolute total power. Steps of the Z-score scale (± 1.5) were adjusted for the “n” in each GDS stage. GDS 1 Z-score step = 0.4. GDS 2 Z-score step = 1. GDS 3 Z-score step = 1. GDS 4 Z-score step = 0.7, and GDS 5 Z-score step = 0.4. To estimate the significance of any regional Z-score for this group’s average data, the Z-score should be multiplied by the square root of the number of participants in the group. For example, for n = 25, the probability associated with an average Z value of 1.25 that corresponds to a standard normal deviation of approximately $6.25\times (1.25\times \sqrt{25})$, that is a probability less than 0.0001.

FU evaluations

Forty-four (81%) of the BL participants converted to a diagnosis within the AD continuum with a mean age of 73 ± 8.5 (range = 52–89) (see Table 2). Normal-functioning older adults at BL (n = 21) developed MCI (GDS 3: 13, 30%) and probable AD (GDS 4–GDS 5: 4, 9%) at FU. Those with BL MCI (n = 20) converted to probable AD (GDS 4–GDS 5: 15, 34%) and severe AD (GDS 6–GDS 7: 4, 9%) at FU. Older adults with BL probable AD (n = 13) converted to severe AD (GDS 6–GDS 7: 8, 18%) at FU. Age was the only demographic that significantly differed across GDS stages (χ² = 14.6, df = 6, P = 0.02). Gender did not significantly differ across GDS stages (χ² = 4.4, df = 6, P = 0.62).

The FU brain images display that only ABS rsEEG theta and REL rsEEG theta significantly and consistently increase across the GDS stages from the frontal to the posterior regions (P ≤ 0.01) of the head (see Figure 3). There were changes in other electrophysiological biomarkers but the changes did not occur across all GDS stages. The MMF rsEEG delta significantly increased only from GDS 2 to GDS 3 in the pre-frontal regions (Fp1, Fp2, F3, F4, Fz; P ≤ 0.01), and only at GDS 4 in the left posterior temporal region (T3, T4). The REL rsEEG alpha significant decreased only from GDS 6 to GDS 7 in the frontal arc (Fp1, Fp2, F3, F4, F7, F8, T3, T4; P ≤ 0.01). The ABS total power significantly increased only at GDS 5 in the entire head (P ≤ 0.01). The rsEEG theta was the only electrophysiological biomarker with a consistent change from normal functioning to severe AD, independent of any significant decrease in REL rsEEG alpha, which occurred only in severe AD.

Display full size

Figure 3. 

Topographic brain images in Global Deterioration Scale (GDS) stages 1–7 at follow-up (FU). MMF: mean frequency; ABS: absolute; REL: relative

Cortical areas with electrophysiological changes correspond to the electrode placements as diagrammed in Figure 1. MMF delta = mean frequency delta, ABS theta = absolute theta, REL theta = relative theta, REL alpha =relative alpha, and ABS total = absolute total power. Steps of the Z-score scale (± 3.0) were adjusted for each GDS stage. GDS 2 Z-score step = 1.2. GDS 3 Z-score step = 1.6. GDS 4 Z-score step = 1.4. GDS 5 Z-score step = 1.2. GDS 6 Z-score step = 1.4. GDS 7 Z-score step = 0.6. To estimate the significance of any regional Z-score for this group’s average data, the Z-score should be multiplied by the square root of the number of participants in the group. For example, for n = 25, the probability associated with an average Z value of 1.25 that corresponds to a standard normal deviation of approximately $6.25\times (1.25\times \sqrt{25})$, that is a probability less than 0.0001.

The figures provide a visual representation of the increasing rsEEG theta occurring consistently in several cortical areas at FU for participants who converted to a diagnosis within the AD continuum. Summarizing these occurrences by gender and age groups may show a pattern for absolute and relative rsEEG theta in the cortical areas.

Absolute rsEEG theta increased for both males and females at GDS 5 as a frontal arc (see Table 3). While the relative rsEEG theta increased for both males and females at GDS 5, females had an increasing rsEEG theta at GDS 6 and GDS 7 as a frontal-lateral arc. The relative rsEEG theta increased in more cortical areas for females who were in the mild to severe stages of AD (GDS 5–GDS 7). The summary indicates more increasing rsEEG theta in females, which can be described as a measure discerning the level of cognitive deficits by gender.

The identification of the increasing rsEEG theta may be more pronounced when grouping the data by age groups. The age groups were created by splitting the ages into four equal groups (see Table 4).

Absolute rsEEG theta increased mostly in participants who were between 66.48 to 80.07 years of age. The increase occurred in age Group 2 (GDS 7) and in age Group 3 (GDS 4 and GDS 5) as a frontal-lateral arc. Relative rsEEG theta increased in age Groups 1–3 (GDS 5) with the increase also occurring in age Group 1 (GDS 4) as a frontal-lateral arc. The increase occurred over the entire head for age Group 2 in GDS 7. For age Group 3, the absolute and relative rsEEG theta had a pattern of increasing activity in GDS 5. This pattern occurred in absolute rsEEG for age Group 2 (GDS 7). It was also evident in relative rsEEG theta for age Group 1 (GDS 4 and GDS 6) and age Group 2 (GDS 5). The increasing rsEEG theta may have been more evident in age Groups 2 and 3. The summary indicates more increasing rsEEG theta in age Groups 2 and 3, which can be described as a measure discerning the level of cognitive deficits by age groups.

This summary is another way to understand the findings. There is a pattern of increasing rsEEG in several cortical areas across GDS 5–GDS 7 for females. This pattern is driven by the increasing rsEEG theta in GDS 5–GDS 7. This pattern is also evident in several cortical areas for age Groups 2 and 3, but it is not specific to absolute rsEEG theta, relative rsEEG theta, or the combination of the two. For certain ages, the pattern is in absolute rsEEG theta and for others in relative rsEEG theta.

Amount of change over time

The amount of change in ABS and REL rsEEG theta was significantly increasing (Z-scores_{AVG CHANGE} = –0.2 to 1.3, P ≤ 0.01) across all GDS stages in the entire head (see Figure 4). For REL rsEEG alpha, a significant decrease only occurred from GDS 4 to GDS 7 (Z-scores_{AVG CHANGE} = –1.1 to 0.5, P ≤ 0.01), indicative of only a change from probable AD to severe AD. The MMF rsEEG delta and ABS total power did not have a significant amount of change across the GDS stages. Only ABS and REL theta had a significant amount of change that demonstrated consistent and increasing changes in brain activity from normal functioning to severe AD.

Display full size

Figure 4. 

The amount of change in the qEEG features in the Global Deterioration Scale (GDS) stages at follow-up (FU). The average (AVG) amount of change in Z-score was computed by taking the summation of the change score (Z-score_FU – Z-score_BL) for each qEEG feature in the selected regions and dividing by the number of qEEG features in the regions. MMF delta = mean frequency delta, ABS theta = absolute theta, REL theta = relative theta, REL alpha = relative alpha, and ABS total power = absolute total power. The regions of interest are the prefrontal region (Fp1, Fp2, F3, F4, Fz) for MMF delta, the entire head (Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T3, T4, T5, T6, Fz, Cz, Pz) for ABS and REL theta, frontal arc (Fp1, Fp2, F3, F4, F7, F8, T3, T4, Fz) for REL alpha, and the entire head for ABS total power. The level of significance was 0.01. qEEG: quantitative electroencephalogram; BL: baseline

Conclusions

An electrophysiological biomarker indicator for AD was detected from the electrophysiological data of normal-functioning older adults who developed a diagnosis within the AD continuum. The indicator emerged as the increasing rsEEG theta progressed across all GDS stages. This increasing rsEEG theta initiated in the frontal region and progressed to the posterior region of the head. This electrophysiological biomarker may enhance the current assessment of AD; nevertheless, further analyses are needed to expound on the clinical utility of this electrophysiological indicator of AD. With further examination, a diagnosis within the AD continuum is not solely dependent on subjective assessment of presenting clinical symptoms. The electrophysiological biomarker will enhance the treatment of persons diagnosed with AD and the identification of interventions to prevent the onset of the illness. Thus, the findings from this study may initiate other studies so that contributions can be made to the implementation of proactive care across GDS stages, transforming the current prognosis from total loss of independence to improved quality of life.