The effect of chronic nausea on gastric slow wave spatiotemporal dynamics in children

Abstract

Background:

Chronic nausea in adolescents with functional gastrointestinal disorders is an increasingly reported but poorly understood symptom that negatively affects quality of life. Functional gastrointestinal disorders are known to correlate closely with slow wave rhythm disturbances. The ability to characterize gastric electrophysiologic perturbations in functional nausea patients could provide potential diagnostic and therapeutic tools for nausea patients.

Methods:

We used high-resolution electrogastrograms (HR-EGG) to measure gastric slow wave parameters in pediatric chronic nausea patients and healthy subjects both pre- and postprandial. We computed the dominant frequency, percentage power distribution, gastric slow wave propagation direction, and speed from HR-EGG.

Key Results:

We observed significant differences in the dominant frequency and power distributed in normal and bradyarrhythmia frequency ranges when comparing patients and healthy subjects. Propagation patterns in healthy subjects were predominantly anterograde, while patients exhibited a variety of abnormalities including retrograde, anterograde, and disrupted patterns. There was a significant difference in the preprandial mean slow wave direction between healthy subjects (222° ± 22°) and patients (103° ± 66°; p ˂ 0.01), although the postprandial mean direction between healthy subjects and patients was similar (p = 0.73). No significant difference in slow wave propagation speed was found between patients and healthy subjects in either pre- (p = 0.21) or postprandial periods (p = 0.75).

Conclusions and Inferences:

The spatiotemporal characterization of gastric slow wave activity using HR-EGG distinguishes symptomatic chronic nausea patients from healthy subjects. This characterization may in turn inform and direct clinical decision-making and lead to further insight into its pathophysiology.

1 |. INTRODUCTION

Estimates of annual adult and pediatric upper gastrointestinal (GI) complaints related to functional GI disorders range in the tens of millions.¹ The myriad of functional GI disorders is characterized by non-specific and overlapping symptomology that often does not correlate with standard diagnostic indicators such as gastric emptying.^2,3 Although structural imaging using CT and ultrasound can identify gross abnormalities and endoscopic biopsies can assess histologic abnormalities, functional gastric disorders frequently occur with no evident structural or ultra-structural abnormalities.

The gastric electrical slow wave mediates neuromuscular interactions in the GI system that determine the functional status of peristalsis and digestion. Disruption of the gastric electrical syncytium by injury or disease interrupts the normal slow wave and produces dysrhythmias, arrhythmias, or uncoupling.^4–7 For this reason, clinical assessment of gastric slow wave parameters is important for the evaluation of gastric pathology. Although high-resolution serosal electromyography and magnetogastrography show that complex slow wave dysrhythmias characterize gastroparesis and chronic unexplained nausea and vomiting in adults,^8–10 there has been little progress in developing objective, reliable, and portable noninvasive tests to characterize chronic nausea or other functional GI disorders in children.

The electrogastrogram (EGG) has historically relied on single-channel or four-channel measurements that can reflect gastric function but is limited to the assessment of temporal parameters like frequency dynamics. Volume-conductor smoothing of intracellular potentials and sequential phase shifts between adjacent cell populations result in a cutaneous potential that resembles a sinusoid.¹¹ Patients with functional gastric disorders have EGG dysrhythmias, but the correlation between dysrhythmias, gastric uncoupling, and delayed gastric emptying is not yet established.^12,13 Many patients have normal EGGs, particularly when only frequency parameters are analyzed,^14,15 spatiotemporal properties may also be important for accurate clinical assessments. Since low-resolution traditional EGG recordings are limited to temporal dynamics, characterizing functional gastric disorders with overlapping symptomology may be enhanced by high-resolution spatiotemporal mapping of slow wave activity. This premise is supported by our preliminary studies.^16,17

Zhou et al.¹⁸ suggested that assessment of slow wave propagation in high-resolution electrogastrogram (HR-EGG) could indicate gastric dysfunction. An encouraging report by Gharibans et al.¹⁹ presented HR-EGG for determining spatiotemporal slow wave characteristics in functional dyspepsia and gastroparesis adult subjects. However, no studies have examined the spatiotemporal gastric slow wave parameters in pediatric patients with chronic functional nausea using HR-EGG. In this study, we hypothesized that HR-EGG is directly correlated with underlying electrical activity, enabling the characterization of spatiotemporal gastric slow wave parameters to distinguish symptomatic pediatric chronic nausea patients from healthy pediatric subjects.

2 |. MATERIALS AND METHODS

All procedures involving human subjects in this study were reviewed and approved by the Institutional Review Board at Vanderbilt University Medical Center. We recruited symptomatic pediatric chronic nausea patients (N = 10; aged 12–17 years; 2 M/8 F) from the Vanderbilt Pediatric Nausea Clinic who have had nausea as their predominant symptom occurring multiple times a week for greater than 3 months, and healthy asymptomatic pediatric subjects (N = 10; aged 11–17 years; 3 M/7 F) from the general population. We did not specify whether the nausea was early morning or postprandial, and patients taking medications were included in the study if they remained symptomatic. Exclusion criteria included a history of known gastrointestinal complications, malignancy, morbid obesity, and pregnancy. Children with vomiting as the primary complaint were also not included in the study. Pediatric chronic nausea patients also underwent physical examination, upper endoscopy, and appropriate radiologic and laboratory investigations to exclude other causes of nausea symptoms. The patient’s current nausea severity was calculated by scoring their perceived level of nausea using the 10-point BARF pictorial scale (0 = absent; 10 = vomiting),²⁰ and their chronic nausea was measured using the five-point Nausea Severity Scale²¹ (NSS) ranging from 0 to 4. In the present study, we included patients with BARF pictorial scale of between two and eight and a total NSS score between 1.7 and 3.6. We obtained written assents and consents from the study subjects and their guardians before enrollment. After a 12 hr fast, we measured HR-EGG using twenty-five silver-silver chloride cutaneous EKG electrodes (Rochester Electro-Medical) arranged in a 5 × 5 square grid centered halfway between the xiphoid and umbilicus, as illustrated in Figure 1. After a 30-minute baseline recording, subjects consumed a protein bar (Gorilla Power Organic, 150 calories, 27 g carbs or ZEGO Vegan, 160 calories, 18 g carbs) and water, and postprandial data were recorded for 1 hr.

FIGURE 1.

Experimental setup to measure HR-EGG with cutaneous electrodes in human subject

To avoid motion artifacts, the EGG recordings were collected in a quiet environment. We advised the participant to remain as still and silent as possible during the recording time. At several times during the recording, we instructed the participant to voluntarily suspend respiration briefly. The respiration-free data improve signal quality by restricting motion artifacts. EGG signals were amplified (BioSEMI Active Two), digitized, and acquired using a custom LabVIEW software (LabVIEW, National Instruments). The HR-EGG signals were originally sampled at 256 Hz but were later down sampled to 30 Hz for analysis. All data analyses were performed using MATLAB (Mathworks). We have included 30 min prepandial and 1-hr postprandial recordings for analysis. We analyzed 120-s-long data segments where segments having excessive noise variance were excluded. Gastric slow wave parameters were calculated for each segment and for each pre- and postprandial period.

For frequency analysis, a bandpass zero-phase second-order Butterworth filter between 1 and 60 cpm was applied to the HR-EGG signals, and the Fast Fourier Transform (FFT) was computed. To further eliminate noise contributions to the HR-EGG signals, we employed an independent component analysis technique (Second Order Blind Identification, SOBI)^22,23 to isolate signal and noise components. We used an automated component selection algorithm to identify components of gastric origin (1–9 cpm) using the following criteria: (1) A gastric component must be primarily sinusoidal, defined as having a correlation coefficient of at least 0.5 with a best-fit sinusoid of matched dominant frequency. (2) A gastric source component must have one dominant peak in the power spectrum. Any extraneous spectral peak with power greater than 50% of the dominant peak classifies the component as multispectral and excludes the component from analysis.²⁴ We reconstructed HR-EGG signals from the chosen SOBI components (SOBI-EGG) by projecting them spatially to the sensor array. Typical high-resolution EGG signals and power spectral densities (PSDs) from a healthy pediatric subject and patient before and after processing with SOBI are shown in Figure 2. We analyzed the dominant frequency and the percentage of power distributed (PPD) in bradygastric- (1–2 cpm), normogasric- (between 2 and 4 cpm), and tachygastric- (4–9 cpm) frequency ranges for all 25 channels and compared these parameters using mean and standard deviation (SD) values. PPD is a measure of the distribution of signal power that could be an indicator of abnormal activity and slow wave uncoupling.^10,25

FIGURE 2.

Butterworth filtered 25 channel HR-EGG and power spectra from (A) healthy pediatric subject and (C) nausea patient. SOBI-reconstructed EGGs and corresponding power spectra in (B) healthy pediatric subject and (D) nausea patient

Pseudocolor spatiotemporal image sequences of reconstructed SOBI-EGG maps allow assessment of slow wave propagation.¹⁰ For the propagation analysis, high intensity noise spikes localized to particular channels and an uneven distribution of signal variances across the array can distort the appearance of spatiotemporal image sequences. Therefore, an alternate processing method was used which first applies a zero-phase second-order Butterworth bandpass filter with a passband between 1 and 20 cpm to the raw EGG signals. We investigated the propagation analysis of EGG data using three SOBI component gastric frequency ranges: 2.5–3.5 cpm, 2–4 cpm, and 1–9 cpm. Using an automated SOBI component selection algorithm, we identified gastric signal components using the following criteria: (1) A gastric component must be primarily sinusoidal, defined as having a correlation coefficient of at least 0.5 with a best-fit sinusoid of matched dominant frequency. (2) A gastric source component must have one dominant peak in the power spectrum. Any extraneous spectral peak with power greater than 50% of the dominant peak classifies the component as multispectral and excludes the component from analysis. For the 2.5–3.5 cpm frequency range, we selected the two gastric components (DFs within a 0.5 cpm range) with the highest correlation coefficients. SOBI components were then isolated and reconstructed onto the sensor array. Next, the reconstructed SOBI-EGG signals were standardized to have a zero mean and unit variance. Following the application of standardization, the SOBI-EGG signals were averaged into a single signal with respect to time and that average signal was subtracted from each signal individually using a technique known as common average referencing. Finally, the processed SOBI-EGG data provide spatiotemporal maps of the EGG signal intensity after placing the signals into a spatial grid and interpolating.

We also estimated the direction and speed of gastric slow waves from HR-EGG using the method described in detail elsewhere.^26,27 This method is briefly summarized below. The Hilbert transform was applied to the processed SOBI-EGG signal at each electrode location to estimate the imaginary part of the analytic signal. The instantaneous phase angle of the analytic signal was calculated for the duration of the timestamp with respect to the positive real axis. The phase angles were unwrapped spatially^28,29 for wavelength calculations and unwrapped over the time dimension for frequency calculations. This phase information was used to calculate the average rate of change of the phase (frequency) and the reciprocal of the absolute magnitude of the phase gradient (wavelength) over the length of the timestamp. Speed was determined by the absolute value of the product of these values at each data point, shown in Equation 1. Propagation direction was determined by the angle of the phase gradient, shown in Equation 2.

$$s(t)=\overline{\left|\frac{\partial \phi (t)}{\partial t}\right|}/\overline{\Vert \nabla \phi (t)\left|\right|}$$ (1)

$$d(t)=ang(-\overline{\nabla \phi (t)})$$ (2)

Phase gradient directionality (PGD) measures the coherency of the signal across the array and is shown in Equation 3. Possible values for PGD are between 0 and 1, where a value of 1 represents a perfectly coordinated signal.

$$\text{PGD}(\text{t})=\Vert \overline{\nabla \phi (\text{t})}\Vert /\overline{\Vert \nabla \phi (\text{t})\Vert }$$ (3)

In our study, a PGD threshold of 0.5 was used to compute the values of wave propagation direction and speed. This threshold helped filter out values of direction and speed where the phase gradients were not well coordinated. This usually occurred at the initiation and termination of each gastric slow wave cycle.

Finally, differences between linear gastric slow wave parameters like frequency, PPD values, and propagation speed were statistically evaluated using Student’s t-test with p-value < 0.05 being considered significant. The results are reported as mean ± SD. For angular data, we used parametric Watson-Williams multi-sample test, a one-way ANOVA test for angular data.

3 |. RESULTS

3.1 |. Frequency analysis

Gastric slow wave activity using HR-EGG was recorded in pediatric nausea patients and healthy pediatric subjects. The mean DF calculated from 25 channel preprandial HR-EGG was 2.8 ± 0.3 cpm for heathy subjects and 2.2 ± 0.2 cpm for nausea patients (p ˂ 0.001), and postprandial DF was 2.9 ± 0.2 cpm for heathy subjects and 2.2 ± 0.3 cpm for nausea patients (p ˂ 0.0001). No significant changes were observed between pre- and postprandial slow wave frequencies in either nausea patients or healthy subjects (see Figure 3a).

FIGURE 3.

(A) Mean slow wave frequency and (B) percent power distributed (PPD) in brady-, normo-, and tachy gastric frequency ranges for HR-EGG in healthy pediatric subjects and nausea patients before and after a test meal. Statistically significant differences between healthy subjects and nausea patients were denoted using *

HR-EGG PPD profiles for healthy subjects and pediatric nausea patients are shown in Figure 3b. Mean PPD showed significant changes in the power distributed in normo- and brady gastric frequencies between nausea patients and healthy subjects in both pre- and postprandial periods. However, there was no significant difference for the tachy gastric frequency range. In particular, the normogastric preprandial PPD was 61.5% ± 4.9% in heathy subjects and 48.3% ± 6.1% in nausea patients (p ˂ 0.001), and the brady gastric PPD was 18.7% ± 5.0% in heathy subjects and 29.4% ± 5.4% in nausea patients (p ˂ 0.001). For the postprandial period, the normogastric PPD was 63.7% ± 5.3% in heathy subjects and 47.1% ± 8.0% (p ˂ 0.001) in nausea patients, and brady gastric PPD was 17.3% ± 3.7% in heathy subjects and 28.3% ± 7.6% in nausea patients (p ˂ 0.01). We observed no significant change between pre- and postprandial PPDs in either nausea patients or healthy subjects.

We also computed the standard deviation (SD) of the DF and the SD of PPD (brady-, normo-, and tachy) across all 25 channels. In both pre- and postprandial periods, the SD of DF and SD of PPD were similar between heathy subjects and nausea patients (Table 1).

TABLE 1.

Summary statistics of gastric slow wave parameters comparing healthy subjects and pediatric chronic nausea patients

	Healthy subjects		Nausea patients		p values
slow wave parameters (Mean ± SD)	Pre	Post	Pre	Post	Pre/Post: healthy subjects	Pre/Post: patients	Healthy subjects/Patients: pre	Healthy subjects/Patients: post
DF (cpm)	2.8 ±0.3	2.9 ± 0.2	2.2 ±0.2	2.2 ±0.3	p = 0.24	p = 0.96	p < 0.001	p < 0.0001
SD of DF across channels (cpm)	0.52 ± 0.1	0.52 ± .0.2	0.54 ± 0.3	0.66 ±0.3	p = 0.97	p = 0.36	p = 0.85	p = 0.26
Brady PPD (%)	18.7 ±5.0	17.3 ± 3.7	29.4 ± 5.4	28.3 ± 7.6	p = 0.51	p = 0.74	p < 0.001	p < 0.01
Normal PPD (%)	61.5 ±4.9	63.7 ±5.3	48.3 ± 6.1	47.1 ± 8.0	p = 0.37	p = 0.72	p < 0.001	p < 0.001
SD of normal PPD across channels (%)	10.7 ± 2.4	10.9 ± 3.5	7.9 ± 3.8	8.8 ±3.0	p = 0.88	p = 0.57	p = 0.09	p = 0.19
Tachy PPD (%)	19.8 ± 6.1	19.0 ± 5.9	22.3 ±4.4	24.6 ± 8.6	p = 0.79	p = 0.49	p = 0.35	p = 0.12
Speed (mm/s) 2.5–3.5 cpm gastric range	13.4 ± 3.0	13.5 ±2.5	11.7 ±2.9	13.0 ±3.8	p = 0.92	p = 0.41	p = 0.21	p = 0.75
Speed (mm/s) 2–4 cpm gastric range	12.0 ± 2.4	11.7 ± 1.6	10.2 ± 3.0	11.0 ±2.4	p = 0.73	p = 0.53	p = 0.16	p = 0.47
Speed (mm/s) 1–9 cpm gastric range	10.7 ± 1.6	10.8 ± 1.5	10.3 ± 3.3	11.0 ±3.4	p = 0.85	p = 0.63	p = 0.75	p = 0.86
Direction (°) 2.5–3.5 cpm gastric range	222 ± 22	163 ± 55	103 ± 66	183 ± 76	p < 0.05	p = 0.21	p < 0.01	p = 0.73
Direction (°) 2–4 cpm gastric range	230 ±39	149 ± 72	206 ± 61	184 ± 62	p = 0.07	p = 0.55	p = 0.43	p = 0.48
Direction (°) 1–9 cpm gastric range	233 ± 41	188 ± 52	223 ± 65	141 ± 60	p = 0.08	p = 0.05	p = 0.77	p = 0.17

3.2 |. Propagation analysis

SOBI-reconstructed spatiotemporal maps from HR-EGG enabled us to observe the propagating slow wave in pediatric healthy subjects and nausea patients. Propagation pattern analysis showed that selecting the two highest threshold gastric components between 2.5 and 3.5 cpm produced propagation maps with clearer resolution when compared to the selection of 2–4 cpm and 1–9 cpm gastric ranges. In most healthy subjects, we found dominant normal anterograde EGG propagation patterns in both pre- and postprandial states for all three gastric ranges. Nausea patients exhibited retrograde, anterograde, disrupted, and a combination of both anterograde and retrograde propagation patterns in both pre- and postprandial periods. Figure 4 shows representative preprandial propagation patterns using the 2.5–3.5 cpm gastric range in a healthy subject and two nausea patients. Anterograde slow wave propagation from the healthy subject is evident in Figure 4a as successive isochronic maps in HR-EGG moving from right to left (corresponding to the subject’s left to right). Anterograde propagation from one nausea patient (Figure 4b), and retrograde and disrupted propagation from another nausea patient taken during two different recording times (Figure 4c,d, respectively) are also shown. Figure 5 shows the corresponding maps for Figure 4a (healthy subject showing anterograde) and 4c (nausea patient showing retrograde propagation) using 2–4 cpm (Figure 5a,b) and 1–9 cpm (Figure 5c,d) gastric frequency ranges.

FIGURE 4.

(A) HR-EGG propagation maps during the preprandial period in a healthy pediatric subject show normal anterograde propagation as the activity moves from right to left (corresponding to the subject’s left to right). HR-EGG nausea patients data show (B) anterograde, (C) retrograde, and (D) disrupted propagation patterns. Position of electrode array is shown in the last map of (A)

FIGURE 5.

HR-EGG propagation maps from a healthy subject and a nausea patient using (A-B) 2–4 cpm gastric frequency range and (C-D) 1–9 cpm gastric frequency range. These maps show corresponding maps for Figure 4A (healthy subject showing anterograde) and 4C (nausea patient showing retrograde propagation). Position of electrode array is shown in the last map of (5A)

We computed the propagation direction and speed of gastric slow wave using the 2.5–3.5 cpm gastric range. The preprandial mean slow wave propagation direction for healthy subjects and nausea patients was 222° ± 22° and 103° ± 66°, respectively. There was a significant difference in the preprandial mean slow wave direction between healthy subjects and nausea patients (p ˂ 0.01). However, the mean postprandial slow wave propagation direction was similar between healthy subjects (163°± 55°) and nausea patients (183° ± 76°; p = 0.73). Control subjects showed a significant change in the mean slow wave direction between pre- and postprandial periods (p ˂ 0.05). However, no such difference was observed in nausea patients (p = 0.21). Figure 6 shows the preprandial representative polar histograms during a 120 second recording time for (a) one healthy subject showing anterograde and three different nausea patients showing (b) anterograde, (c) retrograde, and (d) disrupted propagation patterns. Figure 7 shows polar histograms for the mean gastric slow wave direction from all subjects (a) preprandial heathy subjects; (b) postprandial heathy subjects; (c) preprandial nausea patients; and (d) postprandial nausea patients.

FIGURE 6.

Representative HR-EGG polar histograms showing estimated slow wave directions during a 120 second recording time in the preprandial period for (A) one healthy subject showing anterograde and three different nausea patients showing (B) anterograde, (C) retrograde, and (D) disrupted propagation patterns

FIGURE 7.

Polar histograms for the mean gastric slow wave propagation direction from all subjects (A) preprandial heathy subjects; (B) postprandial heathy subjects; (C) preprandial nausea patients; and (D) postprandial nausea patients

In healthy subjects, the average propagation speed was 13.4 ± 3.0 mm/s preprandial, which did not change significantly in the postprandial period (13.5 ± 2.5 mm/s, p = 0.92). Likewise, in nausea patients, the average preprandial speed was 11.7 ± 2.9 mm/s, which was not significantly different from the postprandial period (13.0 ± 3.8 mm/s, p = 0.41). We found no significant difference in propagation speed between healthy subjects and nausea patients in either pre- or postprandial periods.

When we expanded the criteria for the propagation analysis to include SOBI components in the range of 2–4 cpm, no significant change was observed in the mean slow wave direction between pre- and postprandial periods in either heathy subjects (p = 0.07) or nausea patients (p = 0.55). Likewise, the mean slow wave propagation direction was similar between healthy subjects and nausea patients in both pre- (230° ± 39°, 206° ± 61°; p = 0.43) and postprandial periods (149° ± 72°, 184° ± 62°; p = 0.48). The mean propagation speed was 12.0 ± 2.4 mm/s preprandial and 11.7 ± 1.6 mm/s postprandial in heathy subjects. Nausea patients had a mean propagation speed of 10.2 ± 3.0 mm/s preprandial and 11.0 ± 2.4 mm/s postprandial, which was not significantly different from heathy subjects (pre, p = 0.16; post, p = 0.47).

Likewise, expanding the criteria for component selection in the propagation analysis to 1–9 cpm resulted in no significant difference in mean propagation direction between control subjects and nausea patients in either pre- (233° ± 41°, 223° ± 65°; p = 0.77) or postprandial periods (188° ± 52°, 141° ± 60°; p = 0.17). Similarly, propagation speed changes were not observed to be significant between heathy subjects and patients in either pre- (10.7 ± 1.6 mm/s, 10.3 ± 3.3 cpm; p = 0.75) or postprandial periods (10.8 ± 1.5 mm/s, 11.0 ± 3.4 cpm; p = 0.86). Summary statistics for various slow wave parameters are shown in Table 1.

4 |. DISCUSSION

Research on the clinical utility of the electrogastrogram supposes that cutaneous electrode measurements reflect the underlying gastric electrical activity much like the electrocardiogram reflects conduction abnormalities in the heart. Previous studies showed that standard four-channel EGG exhibits decreased coupling percentages between channels during uncoupling of the gastric musculature which improves the ability of the EGG to detect abnormal slow waves.³⁰ However, the potentially important slow wave parameters of propagation direction and velocity remain elusive in standard EGG recordings. Several pathological gastric conditions have been shown to result in clinically relevant gastric dysrhythmias, but the connection between tachygastria and/or bradygastria reflected in the EGG and true slow wave rhythm disturbances or uncoupling remains unclear.¹³ Although several authors have reported electrical characteristics from standard EGG recordings of gastric electrical activity in healthy children,^31,32 few studies have investigated these in children with functional GI disorders^33,34 and no such studies have been reported in chronic nausea patients using HR-EGG.

Chronic nausea appears to be a heterogeneous condition, but the evidence suggests bidirectional contributions from visceral and central inputs, that is, the brain-gut axis.³⁵ Thus, GI motility disorders, visceral hypersensitivity, and psychosocial factors have emerged as leading candidates in the pathophysiology of chronic nausea.^36,37 Despite the development of motility methods allowing novel exploration of GI physiology, the pathophysiology of chronic nausea in adolescents remains elusive. Functional nausea is a relatively new clinical diagnosis for children, introduced in the most recent Rome IV criteria, and so true prevalence has not been determined. Recent studies report a 45%–53% prevalence of chronic nausea in children with pain-associated functional gastrointestinal disorders³⁷—given that chronic or recurrent abdominal pain affects 8%–25% of otherwise healthy school-aged children, the burden of disease is likely high.³⁸ It is critical to study chronic nausea in children because of its ubiquity, propensity to chronicity into adulthood, deleterious effect on quality of life, and potential for treatment at a time of therapeutic plasticity.^35,39 The use of HR-EGG represents an opportunity to noninvasively assess the potential effects of chronic nausea on gastric slow wave parameters, and ultimately could provide expediated diagnosis and early personalized treatment for effected patients.

In our study, we compare frequency dynamics and spatiotemporal parameters from pediatric nausea patients with healthy pediatric subjects using HR-EGG. We used SOBI to reduce noise contributions from motion artifact, cardiac interference, and respiratory interference while preserving the underlying gastric activity in the EGG signals. We automated the analytic procedure used to compute the frequency, PPD, speed, and propagation direction to eliminate biases caused by investigator interference.

The results of temporal gastric slow wave analysis demonstrate a clear difference in mean DF and normogastric and bradygastric PPDs between healthy subjects and nausea patients in both pre and postprandial periods. However, we did not observe any changes in tachygastria PPD between healthy subjects and nausea patients in either pre or postprandial period, presumably because tachyarrhythmias are isolated local phenomena that may be obscured by the spatial averaging of the HR-EGG. It will be important to study the magnetogastrogram (MGG) to determine the extent to which spatial averaging obscures the potentially important identification of tachygastric frequencies and/or gastric uncoupling in these patients. Another possible reason might be that the true tachyarrhythmia may exist beyond the tachyarrhythmia range we have defined and are exploring. For instance, tachyarrhytmias could be a higher frequency event that surpasses our ability to register due to sampling frequency constraints.

We also did not observe any significant change in SD of DF or SD of PPD across all channels when comparing patients and healthy subjects in both pre-and postprandial state. The lack of difference in SD of DF or PPD across channels might reflect a relative insensitivity of EGG to local electrical disruptions in the gastric slow wave, possibly because of volume conduction effects.⁴⁰

The noninvasive spatiotemporal characterization of gastric slow waves may prove critical to the clinical distinction of functional gastric disorders, and our results suggest that HR-EGG may have a role to play. The increased surface coverage of our 5 × 5 cutaneous electrode grid allows a more precise determination of propagation parameters than is available with standard 4-channel EGG. In the present study, we compared propagation analysis results using three SOBI-EGG gastric frequency ranges. We observed the clearest resolution in EGG propagation maps using a criteria requiring two high threshold SOBI components with dominant frequencies in the 2.5–3.5 cpm range as compared to the expanded ranges of 2–4 cpm or 1–9 cpm. This observation could reflect the potential inclusion of non-gastric sources in the propagation analyses using the expanded frequency ranges. Propagation patterns, which are predominantly anterograde in heathy subjects, appear to be retrograde, anterograde, and disrupted in nausea patients. Using 2.5–3.5 cpm SOBI-EGG selection, the average pre-prandial gastric slow wave direction for healthy pediatric subjects is 222° ± 22° which agrees with the previous reports on slow wave direction estimates in adults.¹⁹ Interestingly, a significant postprandial change in gastric slow wave direction was noted in healthy subjects. Between healthy pediatric subjects and nausea patients, a significant change in the mean slow wave direction was observed during the pre-prandial period but not during the postprandial period. For the 2–4 cpm and 1–9 cpm SOBI-EGG selections, we observed no significant differences in the mean propagation direction between control subjects and nausea patients in either pre- or postprandial periods, presumably since the inclusion of non-gastric noise sources can obscure potential pathognomonic distinctions between nauseated patients and control subjects.

In all three gastric frequency ranges, there was no significant difference in propagation speed between healthy subjects and nausea patients in either pre- or postprandial. Previous studies in our laboratory reported an average propagation velocity of 7.8 mm/s preprandial and 7.4 mm/s postprandial in healthy adults using the magnetogastrogram.¹⁰ Since the electric potential detected in cutaneous recordings is the spatial summation of electric potentials from currents across different gastric regions, it is difficult to determine whether the propagation speed reflects individual slow waves or the summed activity of distinct gastric regions. A lower gastric slow wave speed of 5.8 ± 0.4 mm/s was reported in a recent HR-EGG study in normal adults.¹⁹ Differences in volume conductivity of tissue in pediatric subjects, as well as the possibility of minor spatial aliasing could account for the discrepancy in reported velocities. With the spatial resolution of electrodes that was used, the only effect of aliasing would be an overestimation of propagation speed. Future studies may focus on methods to minimize aliasing effects, such as spatial interpolation or comparison to simulated data that has been subject to aliasing.

A recent study by Ghariban’s et al.¹⁹ used abdominal CT scans to determine the location and geometry of the subject’s stomach in healthy adult subjects, functional dyspepsia patients, and gastroparesis patients. There is a significant amount of anatomical variability between each subject in both the location and shape of the human stomach. Since imaging is not considered part of the standard workup for nausea as a predominant presenting symptom in children, our present study does not incorporate any anatomical comparative analysis between heathy subjects and children with chronic nausea. For future studies, CT imaging during the HR-EGG recording in pediatric subjects would help address these discrepancies and aid the understanding of the inter-subject variability in stomach anatomy, position, and the effects of meals, but the relative risk of radiation exposure should be considered.

Future studies should also investigate the correlation of symptom severity with gastric slow wave parameters, which was confounded in the present study due to the limited sample size. In addition, relationships between gastric electrophysiology and psychological functioning in pediatric chronic nausea patients should also be investigated to better understand the relationship of the brain-gut axis.

It will also be important for future studies to investigate the potential improvement in the characterization of propagation parameters afforded through the use of the MGG and to determine whether and to what extent such improvements offset the additional cost. Previous studies suggested that with its decreased sensitivity to volume conduction effects, MGG could be capable of determining propagation velocity gradients rather than simply estimating a global propagation velocity.⁴⁰ A more detailed characterization of propagation dynamics could lead to greater clinical applicability.

To develop functional imaging techniques using HR-EGG, we are investigating novel feature extraction techniques based on blind source separation and empirical mode decomposition for reducing noise and identifying physiological changes in chronic nausea. Also, we will compare different filter settings and identify the optimized filter setting for gastric slow-wave analysis and interpretation.

In conclusion, our results showed that the noninvasive characterization of gastric slow wave parameters using HR-EGG distinguishes children with functional nausea from heathy subjects, which may allow us to further define clinical phenotype. The increased spatial resolution provided by the HR-EGG and the application of advanced noise reduction techniques may overcome the limitations of standard EGG monitoring observed in clinical practice. Thus, HR-EGG may substantially advance the ability to noninvasively and accurately characterize functional gastric disorders in the pediatric clinic.

Key Points.

• High resolution electrogastrograms (HR-EGG) were measured in ten pediatric chronic nausea patients and ten healthy children.

• Gastric slow wave frequency and percentage power distribution analysis indicated abnormal slow wave activity and slow wave uncoupling.

• HR-EGG propagation patterns in nausea patients were retrograde, anterograde, or disrupted, and the preprandial mean slow wave propagation direction was significantly different.

• The analysis of slow wave activity using HR-EGG represents the first physiologically quantifiable noninvasive assessment method to distinguish pediatric chronic nausea from healthy children.