Assessment of Cognition in Hypertensives and Normotensives: A Comparative P300 Study

Background: Hypertension is an established risk factor for dementia, and the prevalence of hypertension and dementia is rising. Current tests to diagnose cognitive dysfunction at an early stage lack sensitivity and specificity. Recently event-related potentials (ERPs) have gained much attention in diagnosing cognitive dysfunction and are independent of the education status of the subject. This study was done to find any cognitive deficits in the hypertensive population with electrophysiological evidence, which might open the doors for the need to screen the population at an earlier stage so that the population can be prevented from dementia. Methods: Some 31 middle-aged (18-65 years) hypertensives were compared with 31 age, sex, education, and handedness matched normotensives about cognition by neuropsychometric test battery including Hindi Mini-mental Status Examination (HMSE), Hindi Montreal Cognitive Assessment (MoCA), choice reaction time (CRT), and auditory event-related potentials. Results: Hypertensives and normotensives differed significantly concerning P300 potentials’ latency (Fz and Cz P300 latencies: p-value: 0.001), and this change was correlated well with the duration of diastolic blood pressure (BP) (r-value: 0.670). The remaining tests, HMSE, Hindi MoCA, and CRT, were dependent on the education status of the patient. Conclusions: The effect of hypertension on cognitive impairment is evident and can be proved early in its pre-clinical stage using ERPs. Early identification can help in specifying high-risk individuals. ERPs have great potential in screening and diagnosing and can also help in assessing cognition as a reliable tool to show the effect of treatments/interventions on cognitive defects.


Introduction
Hypertension is an established risk factor for cognitive decline, and anti-hypertensives' role in preventing the same further substantiates it [1][2][3]. In 2000, the world was estimated to have nearly one billion people with hypertension, increasing to 1.56 billion by 2025 [4].
The findings of cross-sectional studies of blood pressure (BP) and cognitive function have significantly varied in their results, with some studies having higher rates of cognitive impairment associated with elevated BP, others with low BP, and also U-shaped and J-shaped relationships [5][6][7]. However, longitudinal studies showed a positive correlation between hypertension and cognitive decline. Randomized studies showed a varied picture regarding the effects of treatment of hypertension on cognition [2]. Not only cognitive function but also hypertension is also shown to increase oxidative stress levels and decrease heart rate variability predisposing a person to an increased risk of cardiac morbidity and mortality [8]. The quality of life and sleep were also found to be poor among hypertensive patients [8][9].
After age 65, there is an exponential increase in lacunae and white matter changes [10], which might hinder the detection of early cognitive deficits due to hypertension at a stage when early intervention can prevent further deterioration, as the process of cognitive decline is irreversible. The controversy about the association between later life hypertension and cognitive decline arises because the longitudinal relationship between cognitive change and BP is sensitive to the effects of age, duration of follow-up and number of BP measurements, hypertensive treatment status, comorbidity with cardiovascular diseases and stroke, and possibly subclinical dementia [3].

Auditory P300
Guidelines were practised as guided by Picton et al. [15]. Subjects were informed about the procedure the day before the scheduled ERP test on a phone call and advised for a head bath on the day of testing and to abstain from caffeine and alcohol for 24 h before the test. After explaining the procedure in the laboratory, he/she was seated comfortably, the procedure was explained, and written/informed consent for doing the electrophysiological study was taken from each subject. The skin over the electrode site was cleaned with Nuprep® (Weaver and Company, Aurora, CO) gel and electrodes were attached using Konix® (Turkauz Kong, Instanbul, Turkey) paste. Pure silver disk electrodes (Nihon Kohden, Tokyo, Japan) were used. The electrodes were attached according to the standard protocol, and the test was run after giving the necessary instructions to the patient. Subjects were asked to keep their eyes open as closing eyes can lead to alpha waves in EEG, which can contaminate the response waveform. Odd-ball paradigm was applied in the test, where two types of auditory stimuli (two different frequencies -target and non-target) were given to the subject, and he/she needed to ignore the frequent (non-target) high-frequency tone and respond to the rare low-frequency tone (target) by pressing the right footpad as soon as he/she hears the target stimuli. A trial test for the first 10 stimuli was done on all subjects to familiarize themselves with the test. The detailed characteristics of P300 and its procedure are explained in Appendix 2.

CRT
The CRT was recorded on already validated Deary-Liewald reaction time software on a computer with WindowsTM version 8.1 (Microsoft, Washington). This was designed by Ian J. Deary and programmed by David Liewald, with several iterations between the initial design and the final program was used here. Details of CRT are given in Appendix 5.

Signal Analysis
All the trials of P300 waves were averaged in each session. Peak amplitudes (micro-volts -μV) and latencies (milliseconds -ms) were obtained relative to a pre-stimulus baseline. Data from Cz and Fz electrodes were considered for analysis. All the values were expressed as mean ± SD. A p-value <0.05 is considered significant. All statistics were done with the help of SPSS version 21 (IBM Corp., Armonk, NY). Mean, median, and frequency percentages were calculated for descriptive statistics. An unpaired t-test is applied between hypertensives and normotensives. Similarly, the same test is applied between males and females. Analysis of variance (ANOVA) is applied between age groups and also in education groups. Chi-square is applied to see the association between gender and education. Pearson correlation is applied to see the correlation between the variables. The correlation coefficient is expressed as r-value along with p-value.

Results
In our study, 31 hypertensive subjects were compared with 31 normotensive subjects regarding cognitive parameters. They were analyzed based on age, gender, education, BMI, and BP status, including the duration of hypertension in hypertensives, as given in Table 1.  The HMSE, Hindi MoCA, and CRT scores did not differ between the groups. Cz and Fz P300 latencies differed significantly between the groups, with hypertensives showing longer latencies (p value: 0.001), but their amplitudes did not differ significantly (  The effect of education was compared in all subjects ( Table 3). HMSE and Hindi MoCA scores differed significantly between the education groups ( Figure 2). The effect of education was not significant on ERP parameters (Appendix 6 -- Table 4).

Discussion
The implication of hypertension on the cognitive decline has varied results in the previous studies [5][6][7]. However, all the studies that used ERPs to evaluate cognitive dysfunction proved its effect with electrophysiological evidence [18][19][20][21].
Hypertensives and normotensives did not significantly differ regarding demographic variables in our study. Age, sex, education, and handedness were matched to remove their confounding effect on dependent variables (cognitive function parameters). BMI was significantly higher in the hypertensive group. Except for the BMI and hypertension status, other independent variables (including age, age group, sex, and education level) did not significantly differ between the two groups.
The mean age group of the individuals was 43.29 (±7.84) years in hypertensives and 41.81 (±7.76) years in normotensives. Some 80.6% of individuals were of age 35-54 years. This age group was considered for study since age >65 years itself has more implications on cognitive decline.
Hypertensives and normotensives significantly differed in Cz P300, Fz P300, Fz N100, and Fz N200 latencies. No other parameters, including HMSE scores, Hindi MoCA scores, reaction times in CRT including correct reaction time, reaction time, or wrong attempts, did not differ significantly between the groups. This shows the effect of hypertension on cognition which is represented here by the prolongation of ERP wave latencies. This effect is not evident in HMSE, Hindi MoCA, and CRTs. The increase in P300 latencies is well documented in previous studies [18][19][20][21], and also increase in N200 latencies is shown in studies done in the past by Si et al. and Akhtar et al. [22][23].
Mini-Mental Status Examination has a sensitivity of 18% to detect mild cognitive impairment. Though MoCA has higher sensitivity of 100%, its specificity falls to 82% and has a chance of false positivity. Though MoCA is found to be a reliable screening tool for cognitive impairment, its validity in low-income countries is still questionable, where the literacy rate is low [24]. HMSE and Hindi MoCA are significantly associated with a patient's education level and are not sensitive to recognizing early cognitive deficits. These cognitive deficits are expected to be not very evident in the early years of hypertension to be picked up in HMSE and Hindi MoCA [17].
Several studies done previously interpreted similar results [18][19][20][21]. Few showed a significant decrease in amplitude, but our study did not reproduce such results. The advantage of easy reproducibility and reliability can make this ERP test a standard test to pick up early cognitive deficits [25]. Moreover, it is a non-invasive test that can favor patients' compliance for follow-up studies. The results did not show conflictive evidence compared to previous studies [18][19][20][21]. The association of hypertension with cognitive deficits, though, lacks unanimous conclusions as low BP also proved to be a risk factor for the same [26], but all the studies done with the help of ERP showed a positive association of hypertension with cognitive deficits.
Individuals with all levels of education status, including illiterates to post-graduates, were included in the study. As matching was done, there was no difference between the groups. HMSE and Hindi MoCA were significantly associated with the association of education with dependent variables. There was a strong positive correlation between these scores with education. The association of education with these tests has already been a fact [27]. A similar association was found with CRT, primarily due to the usage of computers.
Hypertension significantly affects cognition, which can be picked up early with the assistance of ERP. The use of MMSE and MoCA scales is primarily dependent on the education level of the subject. Though Hindi versions of both the tests were used in the study, they could not help in detecting early cognitive deficits by eliminating the confounding effect of education. Nevertheless, the use of ERP in detecting cognitive deficits that very early seem to be promising in the near future. Similar conclusions are made by Papaliagkas et al. in a similar study [28]. The effect of hypertension treatment on cognition has also been recently studied, and we have supportive data that treatment of hypertension has decreased the risk of cognitive impairment [29].
Though the treatment of cognitive deficits that are already established is not well known, identifying the risk factors and preventing the dreaded complication is a very effective way of dealing with it and a costeffective solution. The future is promising regarding the development of tools with the help of artificial intelligence to fight cognitive impairment and reverse at least to some extent though it is in its nascent stage right now [30].

Limitations
The additional use of the Pz electrode for studying P300 would have helped in better understanding, and generalized results would have conveyed the reproducibility of findings in all the electrodes. A larger sample size would definitely increase the power of the study, as the prevalence of hypertension in the population is high. A follow-up study would firmly establish the association and causation of hypertension in cognitive impairment.

Conclusions
The effect of hypertension on cognitive impairment is evident and can be objectively proved early in its preclinical stage using event-related potentials. Screening for pre-clinical cognitive deficits can significantly change cognitive disorders' course. Early identification can help specify high-risk individuals and thus can either help us alter modifiable risk factors or guide us in deciding on any intervention for treatment or prevention. As ERPs alone have the potential in the current era to catch very early deficits, they can be considered a game changer in dealing with cognitive disorders. They are beneficial in screening and diagnosing and can help assess cognition as a reliable tool to show the effect of treatments/interventions on cognitive defects. Clinical use of ERPs, which are now being used chiefly as experimental tools, must be revisited, encouraged, and inculcated in practice to prevent cognitive dysfunctions in hypertensive subjects. Early screening and a multimodal non-pharmacological approach focusing on good sleep quality, assessing heart rate variability, and screening for cognitive deficit should be adopted in treating hypertension to prevent the neglected long-term complications of hypertension for a good quality of life.

Appendices Appendix 1
Please see Figure 7 below.

Auditory P300
Guidelines were practised as guided by Picton. Subjects were informed about the procedure the day before the scheduled ERP test on a phone call and advised for a head bath on the day of testing and to abstain from caffeine and alcohol for 24 h before the test. They were allowed to take their routine medication (like antihypertensives in subjects who are already on them). Newly diagnosed cases of hypertension and those hypertensives who are not started on medication till the test are advised to start medication as advised after the test.
After explaining the procedure in the lab (Electrophysiology lab) he/she was seated comfortably, the procedure was explained and informed consent for doing the electrophysiological study was taken from each subject ( Figure 8).

FIGURE 8: Subject performing auditory P300 test.
All subjects were asked to relax for 5 minutes in the lab. No one else except the investigator, patient and patient's attendant was allowed in the lab during the study. A security guard is placed outside the lab and instructed to ensure the same. After explaining the procedure, the silence was maintained except for words of encouragement given by the instructor during the test infrequently. Cell phones of all three members in the room were either switched off or kept in silent mode during the procedure. The skin over the electrode site was cleaned with Nuprep® gel and electrodes were attached using Konix® paste in the electrode placement site (Figure 9). test.
Pure silver disk electrodes (Nihon Kohden) were used. The electrodes were attached according to the standard protocol and the test was run after giving the necessary instructions to the patient. Skin-electrode impedance was reduced to 5KW or less. Subjects were asked to keep their eyes open as the closing of eyes can lead to alpha waves in EEG which can contaminate the response waveform. Odd-ball paradigm was applied in the test, where two types of auditory stimuli (two different frequencies→ targets and nontarget) were given to the subject and he/she needs to ignore the frequent (non-target) high-frequency tone and respond to the rare low-frequency tone (target) by pressing the right footpad as soon as he/she hears the target stimuli. The frequency and probability of the stimuli are described below. The timing and probability of the stimuli were automated in the software. A trial of test for the first 10 stimuli was done on all subjects for familiarizing with the test. A repeat test was also done in a few subjects in view of obtaining better electrophysiological graphs, previously which were having difficulty with marking and interpreting amplitudes and latencies of N100, N200, P200, and P300. A sample ERP is given in Figure 10.
Response times were generated depending on the time taken in milliseconds from stimulus to motor response that is from hearing the target stimulus to pressing the right foot pad. Mean response time, and minimum and maximum response times were given automatically by the software in the results. Missed hits were calculated by the number of missed motor responses.

Rejection criteria for waveforms
• Waveforms with no markable positive or negative peaks and with no significant amplitude (0 uV) for up to three trials • High background noise • >6 missed hits for up to five trials

Appendix 3
Please see Figure 11 below. All the testing was carried out at a fixed time in the morning (9 am-12 pm) to avoid the confounding effect of the circadian rhythm. Four white squares appear in a horizontal line across approximately the middle of the computer screen, set against a blue background. Four keys on a standard computer keyboard correspond to the different squares. The position of the keys corresponds in alignment to the position of the squares on the screen: the 'z' key corresponds to the square on the far left, the 'x' key to the square second from the left, and the 'comma' key to the square second from the right and the 'full-stop' key to the square on the far right. The stimuli to respond appear in a diagonal cross within one of the squares. Participants are instructed to gently rest the index and middle fingers of their left hand on the 'z' and the 'x' keys, and the index and middle fingers of their right hand on the 'comma' and 'full stop' keys. A cross appears randomly in one of the squares and participants are asked to respond as quickly as possible by pressing the corresponding key on the keyboard. Each cross remains on the screen until one of the four keys is pressed, after which it disappears and another cross appears shortly after. The interstimulus intervals range between 1 and 3 seconds and are randomized within these boundaries. The computer program records the response times for each cross, the inter-stimulus interval for each trial, which key was pressed and, in the case of four-choice reaction time, whether the response was correct or wrong, a sample of CRT response is given in Figure 13.

Appendix 6
Please see Table 5 below.  Latencies, potentials, and response time of the ERP waves of both Cz and Fz electrode were compared between the education levels by ANOVA followed by post-hoc Bonferrini tests. In latencies only Fz N200 showed significant difference between the education groups with lower latency periods in higher education categories and longer latencies in lower education groups. But this test when subjected to Bonferrini test did not yield any statistically different results in between the groups. No statistically significant difference was found in the groups between HTs and NTs with regard to potentials.

Dependent variable
Similar to reaction time derived in CRT, response time showed statistically significant difference between the education groups including mean response time and maximum response time, although minimum response time did not show statistically significant different results. All these parameters were slower in less educated subjects. On applying Bonferrini test, the significant difference was found between illiterates and post-graduates in mean and maximum response time, with higher response times in illiterates compared to post-graduate.

Additional Information Disclosures
Human subjects: Consent was obtained or waived by all participants in this study. AIIMS, RISHIKESH IEC issued approval AIIMS/IEC/18/107. The above-mentioned protocol "Assessment of Cognition in Hypertensives and Normotensives-A Comparative P300 Study" was discussed in the Ethics Committee meeting held on 04/01/2018. The Ethics Committee has unanimously approved your protocol. This work will be done under the guidance and supervision of Dr Monika Pathania. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue. Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.