Auditory Cognitive Function Assessment during Pregnancy: An Event-Related Potential and Neuropsychological Study

Objectives: We investigated the auditory cognitive and behavioral functions during 2nd and 3rd trimesters of pregnancy using event-related potentials (ERPs) and different neuropsychology tests, respectively. Methods: The ERPs were studied by using a 128-sensor net, and PAS/COWA, WCST, ZCT, RAVLTIM/RAVLTDR/RAVLTTS, and BDI were tested for neuropsychology assessment. Total 39 subjects were recruited for control group (G1, n=15, non-pregnant), 2nd trimester group (G2, n=12, 13-26 weeks gestation), and 3rd trimester group (G3, n=12, 2640 weeks gestation). Auditory oddball paradigm was used during ERP study. Subjects silently counted only the target stimuli with attention by ignoring standard stimuli. Value of the mean differences of the target and standard stimuli were measured across groups in 10-20 electrode systems. Results: The P50, N100, and P300 ERP components were analyzed. The G3 (at F7 and C3) and G2 (at T4) groups evoked the highest significant amplitudes in P50, and G3 (at Cz and Pz) and G1 (at F8 and T4) evoked the highest significant amplitudes of N100 with significantly prolonged latencies at Cz and O1 (in G1) and F8 (in G2). However, significantly higher amplitudes of P300 were found at Fz (in G3) and FP1 (G2) (G2>G3). On the other hand, neuropsychology tests revealed that G2 possessed the highest significant score in WCST and G1 in ZCT. The G3 performed poorly in all tests. Conclusion: We concluded that pregnant groups performed better in auditory (attention/memory) and executive function assessment. Moreover, the 2nd trimester group has better auditory cognitive function compared to the 3rd trimester group due to the effect of hormonal changes during pregnancy, which might be a positive influence during the pregnancy period.

auditory and visual stimulation/task 9 . Oddball paradigm can catch more neuronal activation during both tasks because participants usually pay more attention during target (odd) stimuli rather than the standard one, and ERP time-locked the responses 10 . The responses of early wave (first 100 ms after stimulation) and late wave (after 100 ms after stimulation) can be recorded during ERP study. The P25, N20, P50, and N100 ERP components are the early components, and P200, N200, P300, N400, and P600 are the late components. 'P' indicates positive peak and 'N' is negative peak for amplitudes. On the other hand, the number after 'P' and 'N' indicates the latencies after stimulation 11 . In this study, we analyzed two early ERP components of P50, N100, and one late component of P300 by using auditory oddball paradigm.

P50, N100, and P300 of ERP components:
The sequences of ERP components vary in latencies, amplitudes, and scalp distributions 12 . The P50 and N100 are exogenous attributes as they are influenced by the physical characteristics of stimuli, independent of cognitive factors, while P300 is an endogenous attribute. The endogenous attributes do not depend on physical stimuli and get influenced by cognitive factors, like attention 13 . Based on the human and animal studies, the P50 ERP component is largely generated by the hippocampus area by providing preattention gating to repeated stimulation 14 . Sensory gating protects the brain by preventing overflow of redundant repetitive stimulus 15 , thus, explaining the mean difference in peak values between target stimuli and deviant stimuli in oddball paradigm task. Abnormalities of sensory gating are hypothesized as one of the indicators in cognitive impairment 14 , like schizophrenia 16 . The auditory N100 is an exogenous component, and it depends on the pitch of the auditory stimulus 17,18 . The N100 is also known as the 'vertex potential' as it has maximum amplitude over Cz compared to other sites 15 . The P300 ERP component has two sub-components, P3a and P3b, that differ by the peak latency. P3a has more frontocentral distribution compared to P3b 12 . However, the polarity, scalp distribution, and latency did not solely reflect the origin of a component; instead, it depends on the difficulty of the discrimination task 18 .

Relationship of amplitudes and latencies of ERP components with cognitive function:
Based on previous reports, the interpretation of the amplitudes and latencies of P50 and N100 are variable. The increase in amplitude of N100 and longer latencies were found in low cognitive function in bipolar disorder patients 19 . Schizophrenia patients displayed poor attention, with a smaller amplitude of the N100 component. It is because the small amplitude is related to the low attention to the stimuli. However, a few studies followed different methods in interpreting the amplitude and latencies of these early components. There is a common misinterpretation about early exogenous ERP components, as some researchers did not consider the refractory short inter-stimulus interval duration. Thus, the early ERP components, such as P50 and N100, tend to have small amplitude when the elicited auditory stimulus follows soon after another stimulus 20 .This means that the early ERP component was expected to have small amplitude when there was low sensory and perceptual function. Thus, we adapted these studies in our P50 and N100 data interpretation.
Lower amplitudes and longer latencies of P300 ERP component are considered as lack of attention, which is also called mild cognitive impairment. The generators of P300 components are from the frontal, temporal and parietal areas, and these areas are easily evoked by auditory oddball task, where participants can focus more during odd stimuli. On the other hand, difficult task can diminish these areas 21 .

Neuropsychology tests and pregnancy:
Few studies were conducted to assess memory and attention during pregnancy, mainly in the 3 rd trimester, with different neuropsychology tests. Memory testing during pregnancy was done previously 22 , especially to determine memory impairment during last trimester 23 . Some researchers investigated attention during pregnancy, where the results were controversial with attention deficit 24 and no changes of attention 25 . Some general memory and attention tasks were performed in the previous studies to assess attention. Digit span tests (digit span forward: DSF and digit span backward: DSB), resulting during high-risk pregnancy, showed a significant decline in this group compared to the normal uncomplicated pregnancies during DSF condition but not in DSB condition 26 .
The different trimesters of pregnancy can make different impacts on cognitive functions 23,27 . Some researchers found no difference in the cognitive changes across all trimesters of pregnancy 28 . Therefore, a new study is needed to investigate the changes in cognitive function not only at mid trimester but also in the third trimester so that the changes can be further studied and compared.

Neuropsychological tests for assessing cognitive function
The results of neuropsychological tests among pregnant women are different. The Wisconsin Card Sorting Test (WCST) was widely used to assess cognitive disorder. Lower WCST score was found in schizophrenic patients who have frontal lobe dysfunction 29 . Studies reported that WCST can assess abstract thinking 29 , abstract reasoning and concept formation 30 , and also the function of the frontal lobe 29,30 . Pregnant women may have lapse of memory due to cognitive concerns during pregnancy 31 , and they are prone to fatigue and sleep disturbances 6 . They also tend to have depression and negative effect. Depression or psychological stress has an impact on the memory of pregnant women 6 . The Zazzo's Cancellation Test (ZCT) is able to measure mild cognitive impairment 32 . In order to assess and measure the cognitive abilities, it was suggested to combine multiple types of neuropsychological tests in the study because different tests have different capabilities of clinical diagnosis. Therefore, seven different types of neuropsychological tests were used to assess cognitive dysfunction in the preclinical phase of dementia. The tests were Mini Mental Status Examination, ZCT, Digit symbol Substitution Test, Benton Visual Retention Test, Wechsler Paired Associates Test, Isaacs Set Test, and Weschler Similarities Test 33 .

Significance of this study:
There were mixed and inconsistent findings in the cognitive functions among pregnant women, in which the neuropsychological tests and self-report were not related significantly. Women frequently selfreport worse memory during pregnancy and are more likely to have memory complaints than non-pregnant women 31 . However, the self-report is not reliable in clinical diagnosis and identification. Therefore, a combination method of investigation is needed to investigate the cognitive functions among pregnant women. This recent study was done to identify the functions through a combination of objective test (ERP) and subjective tests (neuropsychological tests). On the other hand, the cognitive functions of auditory attention and memory between each trimester can be investigated. As mentioned in the previous literature, late stages of pregnancy tend to have hormonal fluctuations. The progesterone level is higher in pregnant women for the maintenance of the endometrium during early pregnancy, which later grows into a fetus. Besides, the regulation of the hormones allowed for the preparation for child labor in the third trimester stage 34 . Mild deficits of late pregnancy (i.e., third trimester) in attention, verbal, and spatial memory were associated with the fluctuations of estrogen 35 . Hence, through this study, it was worthwhile to know the exact neuronal network processing for therapeutic/rehabilitation purpose to improve the auditory cognition among pregnant women.
The aim of this study is to investigate the amplitudes and the latencies of P50, N100, and P300 ERP components and different neuropsychology tests among subjects with 2 nd and 3 rd trimesters of pregnancy compared to the normal healthy subjects by using the auditory oddball task.

Study design:
We received human ethical approval from the Ethical Committee of USM (USM/JEPeM/15090294) before starting the experiment. All subjects were recruited through personal communication, internet, and notice board advertisement. This study was conducted in the MEG/ERP laboratory at Hospital UniversitiSains Malaysia (HUSM) by using a 128-electrode sensor net. Written informed consent was obtained from all subjects prior to the experiment. The subjects underwent ERP experiment and neuropsychological tests and were assessed by a clinical neuropsychologist.

Study populations and groups:
The study population was calculated by a statistician using Power and Sample Size software. There were three groups: control healthy subjects (nonpregnant, G1), 2 nd trimester pregnant group (G2), and 3 rd trimester pregnant group (G3). The numbers of subjects were, G1=15, G2=12 and G3=12. We did not add the 1 st trimester of pregnancy due to vulnerability.
We excluded the subjects suffering from major diseases like hypertension, diabetes, and kidney diseases 36,37 and who had a history of drug and alcohol abuse 38 as these factors can affect the cognitive functions. All women in these three groups were age-matched (20-40 years) 39 and educationmatched (higher than SijilPelajaranTinggi Malaysia; Malaysian education system) as lower education reflected lower cognitive function like attention 40 .

Event-Related Potential: ERP Recording
The electrolyte solution was made using 11 mg potassium chloride (KCI), 5 ml Johnson Baby shampoo, and 1 L of distilled water. The 128 ERP sensor net/cap was soaked in this solution for 10 min. The sponge of the electrode in the net absorbed the electrolyte solution and helped to reduce impedance during recording. Impedance was set at <50 KΩ.

ERP procedure (Experimental paradigm)
E-Prime v 2.0 software (Psychology Software Tools, Inc, Sharpsburg, Pennsylvania, USA) was used for the presentation of stimuli, timing operations, and data collection. Each subject sat in a dimly lit room with headphone placed on both ears, wearing the 128-electrode sensor net. In the auditory oddball paradigm, all participants counted target tones (60 dB sound pressure level (SPL), which was low frequency (20%) and high pitched (2000 Hz)) silently while ignoring standard tones (60 dB SPL, high frequency (80%) and low pitched (1000 Hz)). Tone duration was 100 ms with a rise/fall time of 10 ms. All data were recorded on Net-Station software 5.2 (Eletrical Geodesics, Inc, Eugene, OR, USA). The ERP component of latencies and amplitudes of P50, N100, and P300 were analyzed.

Neuropsychological tests
Subjects were assessed based on different types of neuropsychological tests, such as Controlled Oral Word Association (COWA or PAS), WCST, ZCT, Rey Auditory Verbal and Learning Test (RAVLT), and Beck Depression Inventory (BDI).

Controlled Oral Word Association
The COWA or PAS test is also known as the "Verbal Fluency Test". It consists of the task of naming a word for the alphabets P, A, and S. The three set of letters made it known as the PAS test 41 . This test requires subjects to mention verbally as many words as possible starting with letters P, A, and S in one minute based on the category of the words, such as animals, fruits, and furniture and excluding proper names and numbers 42 . The sum of correct words was scored and adjusted based on age, sex, and education. Any word errors or repetitions were also calculated for the scoring.

Wisconsin Card Sorting Test
The WCST can assess the executive function 43 , and it is a specific test for frontal lobe dysfunctions 29,43 . Four different cards are placed before the subjects. The cards have different features of color (i.e., red, green, blue, or yellow), form (i.e., circles, stars, squares, or crosses), and number (i.e., 1, 2, 3, 4). The subjects need to sort and match the cards according to the specific features and goals, as instructed by the researcher during the test. Throughout the test, the matching rules are changed after 10 successive matchings. Meanwhile, the researcher gives out scores for each error and correct matching in the test. In this test, the error rate is the main score for the subjects/patients.

Zazzo's Cancellation Test
This is a visuospatial perception test, including selective attention and response speed to particular tasks 44,45 . The subjects are required to cross out particular target sign on the sheet of white paper containing a number of signs. Then, the scores are calculated based on the number of signs that were correctly crossed out (ZCTtot) and the time taken to complete the task (ZCTtime). Impaired performance signifies lower scores.

Rey Auditory Verbal and Learning Test
The RAVLT is a well-known test to evaluate the ability of encoding, combining, storing, and recovering verbal information at different stages of immediate memory 46 . During this test, the researcher will read out 15 noun words from list A to the subjects at the timing rate of one word per second. Then, the subjects are required to recall as many words as possible. The procedure is repeated for 5 trials and the scores are averaged into one score based on the 5 trials. Then, 15 new noun words from list B are presented to the subjects, which acts as interference. Then, the subjects need to recall the maximum possible nouns from list A. This step is carried out to evaluate the delayed recall, immediate recall, and long recall of memory 46,47 .

Beck Depression Inventory
The BDI intended to determine the severity and intensity of depression symptoms 48 . The evaluation is based on the rating scale of 21 specific items, and the scale ranges from 0 to 3. Individuals with a clinical diagnosis of depression will be assessed and scored differently than the general population. For the general population, a score of 21 and above is associated with depression while for those clinically diagnosed with depression have 3 stages of scores. Those with 0 to 9 represent minimal depressive symptoms, 10 to 16 is mild depression, scores of 17 to 29 indicate moderate depression, and severe depression scores range from 30 to 63. The benefit of this test is that majority of people can finish it in five to ten minutes as long as the environment is suitable for the subjects to concentrate 48 .

Data analysis:
The ERP data were analyzed using EGI Net Station 5.3 Software. Filtering was 0.3-50 Hz with 0.5 as the stimulus rate. Segmentation was 100 ms before the stimulation and 1000 ms after the stimulation. Artifacts, such as eye blink, eye movements, and movement artifacts were removed using the artifact detection tools in Net-Station Software. The artifacts were removed by notch filter setting that filtered out eye blink (> 200 µV), eye movement (70 µV), and bad channels (>400 µV). Baseline was corrected to 100 ms before the stimulation 49,50 .
The amplitudes and latencies of P50, N100, and P300 ERP components on 19 channels of electrodes were extracted from Net Software Station. To know the significant value among groups, data from ERP and neuropsychology tests were analyzed using Statistical Package for the Social Sciences 22 (SPSS22) software, using one-way ANOVA analysis. The significance level, p-value, was set as p ≤0.05. Further, ANOVA post hoc Bonferroni test was conducted to identify the mean differences between groups 49,50 .
Ethical clearance: This research study was approved by ethics committee of School of Medical Sciences, Universiti Sains Malaysia, Kubang Kerian 16150, Kota Bharu, Kelantan, Malaysia.

Results of ERP components:
Nineteen electrode channels were analyzed in 10-20 systems. The values of the mean differences of standard and target stimuli for the three groups were measured for the amplitudes and latencies of the P50, N100, and P300 ERP components. Only the significant changes at some electrodes were found for all three ERP components (Table 1). Grand average waveforms of the P50, N100, and P300 ERP components for G1, G2, and G3 groups were shown in Figure 1.The waveforms were shown only at the significant electrode sites. Three electrodes F7, C3, and T4 showed significant differences among groups for the amplitudes of the P50 ERP component. The G3 group evoked the highest amplitudes at F7 (G3> G1 >G2) (p=0.034), C3 (G3> G1 >G2) (p=0.034) locations. On the other hand, G2 group induced the highest amplitude at T4 (G2> G3 >G1) (p=0.031) electrode site. The group effects at F7, C3, and T4 were F(df)= 3.660(2,36), F(df)= 3.679(2,36) and F(df)= 3.797(2,36), respectively. There were no significant differences for the latencies among groups for P50 component (Table 1). group. There were no significant differences in P300 latencies among groups ( Table 1).

Results of neuropsychology Tests:
The scores of neuropsychological tests among groups were given in Table 2. WCST and ZCT showed significant differences among groups. G2 group revealed the highest significant scores in WCST (p=0.001), followed by the scores in ZCT test. G1 group had the highest significant scores in ZCT (p=0.001). G2 group revealed highest scores in ZCT, but lowest scores in WCST tests. The group effects for WCST wereF(df)=843.776(2,36) and for ZCT were F(df)=528.598 (2,36). No significant difference was found in PAS, RAVLT, and BDI test scores among groups (Table 2).

Discussion:
The aim of this study was to investigate the cognitive functions, like auditory attention and memory among pregnant women in the 2 nd (G2) and 3 rd trimesters (G3) of pregnancy by comparing them with non-pregnant women (G1), the control group. Amplitudes and latencies of the P50, N100, and P300 ERP components were analyzed among groups using auditory oddball task in ERP study. The ERP components of P50, N100, and P300 are the reflection of sensory gating, auditory perception, and attention, respectively. We assessed the executive function and auditory memory function using five different tests such as WCST, ZCT, RAVLT, COWA or PAS/FAS, and BDI. The 3 rd trimester group (G3) evoked significantly highest amplitudes for P50 (at two areas), N100 (at two areas) and P300 (one area) ERP components. The 2 nd trimester group (G2) possessed significantly highest amplitudes for the P50 (at one site) and P300 (at one site) and also significantly prolonged latency at one site for N100 ERP component among groups. However, control (G1) group evoked significantly highest amplitudes at two sites and prolonged latencies at two sites also. G2 group had the highest significant scores in WCST, whereas G1 group had the highest ZCT scores. On the other hand, G3 group revealed poor scores in both WCST and ZCT.

P50 ERP component
The interpretation of amplitudes and latencies of both the P50 and N100 ERP components is opposite from the P300 component. As mentioned, higher amplitude reflects poor performance in case of P50 and N100 components, but in the case of P300, it reflects higher performance 51 .
There have been no studies on the P50 ERP components of the pregnant group. Few studies were conducted only to get more information about this component. The P50 ERP component conveys the flow of auditory information from the thalamus to the auditory cortex of the brain, which is important for the detection of sensory gating process of the stimuli 52 . Moreover, this component is the preattentive arousal of stimuli 53 . Higher amplitudes of P50 ERP component were found in Alzheimer's patients, diagnosed as a mild cognitive impairment 51 .
In our study, we found that the 3 rd trimester pregnant group (G3) had the highest significant amplitudes in two electrodes (F7 and C3) and the 2 nd trimester pregnant group (G2) had the highest significant amplitude of P50 component in one area (T4) ( Table  1). There were no significant differences in latencies among groups. Therefore, based on the results of P50 amplitudes and taking the result from Golob et al., 51 , we assume that the 3 rd trimester pregnant group has mild cognitive impairment and poor sensory gating.

N100 ERP component
The interpretation of the amplitudes of N100 is the same as for the P50 ERP component, but no clear interpretation has been made for N100 latency. The N100 ERP component reflects the auditory perception. Higher amplitude of the N100 component reflected the poor developmental perception of the auditory system 54 . One study found a higher amplitude of the N100 component in alexithymic subjects, and they interpreted that they have impairment of auditory perception 55 . In our study, the 3 rd trimester pregnant group revealed highest significant amplitudes in two electrodes (Cz and Pz), and the control group had the highest significant amplitudes in another two electrodes (F8 and T4). The longest significant latencies were found in the control group in two areas (Cz and O1), and one area (F8) reflected significant longer latency in the 2 nd trimester group (Table 1). Taking the points from the previous interpretations 54, 55 , we can say that the 3 rd trimester pregnant group has poor auditory perception.

P300 ERP component
Studies have proved that the P300 ERP component is the marker to diagnose cognitive impairment 56,57,58 .
Lower amplitudes and longer latencies of the P300 component indicated mild cognitive impairment in the spinocerebellar ataxia type 2 (SCA2) patient groups 59 , in old ages 60 , and in Alzheimer diseases 61 .In our study, we found that the pregnant group evoked the highest significant amplitudes at two electrodes, Fz and FP1, compared to the control group. Within pregnant groups, the 2 nd trimester pregnant group evoked higher (8.28±4.52 µV) P300 amplitude at FP1 area compared to the 3 rd trimester pregnant group at Fz area (5.87±2.73µV) (Table 1). Therefore, we interpreted that during the 2 nd trimester, pregnant women have better cognitive function/attention compared to the 3 rd trimester and control group. A little shorter amplitude of P300 in the 3 rd trimester pregnant group suggested that attention might increase during pregnancy but can be reduced during the 3 rd trimester nearer to childbirth 27,56 .

Neuropsychological tests
To Here, we performed WCST, ZCT, PAS, RAVLTim, RAVLTdr, RAVLTts, and BDI neuropsychological tests in pregnant (2 nd and 3 rd trimesters) and control groups. Both WCST and ZCT showed significant differences among groups and explored the executive function. The highest significant WCST scores were found in the 2 nd trimester group. Control group possessed the highest significant ZCT scores among groups. The 3 rd trimester pregnant group scored poorer in all tests compared with the 2 nd trimester group ( Table 2). The poor scores in the neuropsychology tests may be due to the hormonal effect 34 . We clarified the scores of the neuropsychology tests among the groups in our study. In our study, we assumed that the pregnancy hormone progesterone is responsible for a better executive function and auditory memory/ attention in the 2 nd and 3 rd trimesters of pregnancy, which might be reduced in the 3 rd trimester of pregnancy.

Conclusion:
We studied the auditory cognitive function and behavioral tests among pregnant women using the ERP study and neuropsychology tests. The 2 nd trimester pregnancy group showed better auditory attention than the 3 rd trimester pregnant group.
Although the 3 rd trimester pregnant group was advocated to have mild cognitive impairment compared to the 2 nd trimester group, it does not significantly affect the daily life of our subjects as the 2 nd trimester pregnancy group still has a better auditory cognitive function and executive functions compared to the control group. The mild cognitive impairment found among the 3 rd trimester group in this study may benefit psychologists, neurologists, and obstetrics to plan rehabilitation and assessment for future mothers in order to help them improve their cognitive function throughout the pregnancy.

Limitations:
1. The sample size of this study was small. We need a large sample size to get more reliable results.
2. The 1 st trimester pregnancy group is not included as this trimester is vulnerable.

Authorship Contribution
TB and FR performed data acquisition, analyses, and interpretation and wrote the manuscript.

Disclosure/Acknowledgment:
This writing is a technical report on the basis of the grant proposal, which was supported by the Universiti Sains Malaysia (USM) short-term grant (304/ PPSP/61313160). A part of this article was presented in a conference and published as an abstract in the Malaysian Society of Neurosciences (MSN) 2017 (myneuro2017). Moreover, one master's student's thesis was written on the basis of this proposal, and the thesis is also converted into a monogram report was not published elsewhere. The authors confirm that they have no conflicts of interest.