Noninvasive magnetogastrography detects erythromycin-induced effects on the gastric slow wave

Abstract

Objective:

The prokinetic action of erythromycin is clinically useful in conditions associated with gastrointestinal hypomotility. Although erythromycin is known to affect the electrogastrogram, no studies have examined the effects that erythromycin has on gastric slow wave magnetic fields.

Methods:

In this study, gastric slow wave activity was assessed simultaneously using non-invasive magnetogastrogram, electrogastrogram, and mucosal electromyogram recordings. Recordings were obtained for 30 minutes prior to and 60 minutes after intravenous administration of erythromycin at dosages of 3 mg/kg and 6 mg/kg.

Results:

Magnetogastrogram (MGG) recordings showed significant changes in the percentage power distribution of gastric signal after infusion of both 3 mg/kg and 6 mg/kg erythromycin at t= 1-5 minutes that persisted for t = 30-40 minutes after infusion. These changes agree with changes observed in the electromyogram. We did not observe any statistically significant difference in MGG amplitude before or after injection of either 3 mg/kg or 6 mg/kg erythromycin. Both 3 mg/kg and 6 mg/kg erythromycin infusion showed retrograde propagation with a statistically significant decrease in slow wave propagation velocity 11-20 min after infusion. Propagation velocity started returning towards baseline values after approximately 21-30 min for the 3 mg/kg dosage and after 31-40 min for a dosage of 6 mg/kg.

Conclusion:

Our results showed that the magnetic signatures were sensitive to disruptions in normal slow wave activity induced by pharmacological and prokinetic agents such as erythromycin.

Significance:

This study shows that repeatable noninvasive bio-electro-magnetic techniques can objectively characterize gastric dysrhythmias and may quantify treatment efficacy in patients with functional gastric disorders.

I. Introduction

Erythromycin is a motilin receptor agonist with a profound effect on gastrointestinal motor activity. Erythromycin is used clinically as a prokinetic to treat gastrointestinal hypomotility in conditions such as diabetic gastroparesis and idiopathic pseudo-obstruction [1]. Erythromycin induces phase III activity of the interdigestive migrating motor complex (MMC) in humans [2] and can improve gastric emptying in gastroparesis patients [3, 4]. Previous studies to evaluate the dose-response curve for erythromycin in healthy subjects suggested that intravenous administration of 3 mg/kg erythromycin was associated with statistically significant increase in liquid gastric empting compared with placebo [5]. Previous research on the therapeutic effect of erythromycin in patients with diabetic gastroparesis showed that erythromycin reduced the prolonged gastric-emptying times for both liquids and solids to normal [3]. Erythromycin has also shown potential benefit in other areas, including patients after vagotomy and patients with chronic intestinal pseudo-obstruction [6, 7].

Although several studies have reported the effect of erythromycin on GI motility, the effect of erythromycin on gastric slow wave activity has been poorly characterized to date, despite considerable research significance. Electrogastrogram (EGG) irregularities, including bradygastria and signal amplitude increase after intravenous erythromycin infusion (6 mg/kg), have been reported in normal human subjects using traditional low channel cutaneous EGG recording [8]. While the traditional EGG using a limited number of surface electrodes shows temporal features of the electrical activity of the stomach, its sensitivity to the electrical conductivity profile of the abdomen complicates the routine analysis of other clinically relevant parameters such as propagation pattern and propagation velocity [9, 10]. A recent study by Gharibans et al. addressed the spatial limitations of traditional EGG by presenting high-resolution EGG for determining gastric slow wave propagation characteristics [11]. This study, however, does not include obese subjects, and it is possible that the low-conductivity fat layers in the abdomen will present a challenge in detecting gastric slow wave propagation.

The magnetogastrogram (MGG) overcomes many of the inherent limitations of the traditional EGG [12–15]. Magnetic fields generated by gastric slow wave currents are less influenced by tissue conductivity differences than are electric potentials, and thus facilitate a more complete characterization of clinically relevant spatiotemporal slow wave parameters [16, 17]. Studies in our lab showed abnormal slow wave spatiotemporal propagation parameters measured using MGG in patients with gastroparesis [18]. Similar slow wave propagation pattern abnormalities associated with gastroparesis have been observed in serosal recordings [19].

No studies have examined the effects of erythromycin on the gastric slow wave magnetic fields measured by MGG. We hypothesized that MGG recordings reflect gastric slow wave changes caused by intravenous erythromycin infusion in normal human subjects.

II. PROCEDURE

Ten healthy human volunteers (8 men and 2 women, BMI range 21 to 29.5, age range 20 to 38) participated in this study. All subjects were free of any gastrointestinal symptoms; had no history of diabetes, gastrointestinal illness, or surgery, and none were on medication known to alter gastrointestinal motor or electrical activity, and female subjects were not pregnant. All the experimental procedures were reviewed and approved by Vanderbilt University Institutional Review Board and General Clinical Research Center, and informed consent was obtained from each subject.

Subjects underwent an eight hour overnight fast prior to the study. Multichannel MGG and EGG were acquired simultaneously with an eight channel mucosal electromyogram (EMG). For the EMG recordings, we utilized a custom-fabricated eight channel mucosal electrode array integrated in a 4.5 mm diameter naso-gastric (NG) catheter (Mui Scientific, Ontario, Canada). Electrodes were platinum rings installed concentric on the catheter. The 4 mm long electrodes were spaced every 15 mm along the NG catheter resulting in a recording length of 13 cm. EGG was measured using 16 silver-silver chloride cutaneous EKG electrodes (Rochester Electro-med, Rochester, MN) arranged in a 4×4 array on the abdomen above the epigastrium as shown in Fig 1a. A multichannel Superconducting Quantum Interference Device (SQUID) magnetometer (model 637; Tristan Technologies, CA, USA) was used to measure MGG. The SQUID converts magnetic flux incident on detection coils into voltage signals. The magnetometer consists of 19 normal component sensors in a hexagonal close-packed array that measure magnetic fields perpendicular to the body surface. At the time of recordings, two of the 19 normal component sensors were not operational and data from these sensors were not included in the analyses.

Figure 1.

(a): Experimental setup to measure MGG using a SQUID magnetometer, EGG with cutaneous electrodes, and EMG with NG electrode catheter in normal controls. The SQUID sensor array shown in the inset consists of 19 normal-component (z) sensors and five vector channels (marked with x) that also sample x and y magnetic field components (18) ; 1(b) : X-ray showing the placement of NG tube in the stomach and the location of 8 electrodes (15).

The NG tube was positioned in the pyloro-antral region along the greater curvature of the stomach with location verified by X-ray (see Fig 1b) [15]. Volunteers lay supine underneath the SQUID magnetometer in a magnetically shielded room. Simultaneous MGG, EGG and EMG slow wave signals were obtained for a 30 minute baseline recording. Erythromycin was then infused intravenously for 20 minutes at a dosage of either 3 mg/kg (N=5; 4 men and 1 woman) or 6 mg/kg (N=5; 4 men and 1 woman). Recordings were obtained for a period of 60 minutes following erythromycin infusion.

Electrical signals were acquired with a sample frequency of 256 Hz using an Active Two amplifier system (BioSEMI, Amsterdam) with a 24-bit ADC on a personal computer running LabVIEW (National Instruments. Austin, TX). SQUID signals were collected with a hardware-imposed sampling rate of 3 kHz using a LabVIEW-based data acquisition system. All data were synchronized and downsampled to 30 Hz for analysis. The recorded signals were transferred into MATLAB (Mathworks, Natick, MA, USA) and filtered using a second-order Butterworth filter with a bandpass of 1 - 60 cycles per minute (cpm). Frequency spectra were computed using Fast Fourier transform (FFT) in MATLAB. To further reduce noise in MGG and EGG signals, a second order blind identification (SOBI) signal processing algorithm was employed [20, 21]. We used an automated component selection algorithm to select primary sinusoidal gastric signals (1-9 cpm) using the following criteria

• a correlation coefficient of at least 0.5 with the best-fit sinusoid at a frequency equal to the dominant frequency of the SOBI source;

• no spurious or additional spectral peak with power greater than 50% of the dominant peak.

SOBI components corresponding to respiratory, cardiac, and motion artifact were eliminated and the remaining dominant gastric signal components were reconstructed onto the sensor array, allowing us to localize gastric sources. In our notation, SOBI-MGG and SOBI-EGG refers to the filtered MGG/EGG signals processed using the SOBI algorithm to identify gastric signal components. We used the central channel of the reconstructed SOBI-MGG/SOBI-EGG array for computations of the dominant gastric slow wave frequency [18]. The central channel is chosen because of its presumed proximity to the subject’s stomach. The central channel of the SOBI-reconstructed MGG or EGG contains weighted gastric signal contributions from the entire recording array.

We determined frequency, amplitude, and the percentage of power distributed (PPD) in ranges normally considered bradygastric (1-2 cpm), normogastric (between 2 and 4 cpm), and tachygastric (between 4 and 9 cpm) from the gastric slow wave measured by MGG, EGG and EMG. The inability to detect consistent phase shifts from EMG and EGG electrodes prevented the computation of propagation velocity. However, we identified propagation patterns and computed the propagation velocities in SOBI-MGG signals. We reconstructed the data using those components in the sensor array and prepared spatiotemporal maps of signal intensity. Tracking the signal intensity maximum of the resulting maps over time allows an estimation of the gastric slow wave propagation velocity. For the propagation velocity estimate, we averaged the velocity obtained from three successive slow wave cycles [16, 18, 22]. Furthermore, we investigated the correlation of EMG signals with corresponding MGG and EGG components isolated with SOBI. The results are expressed as mean ± standard error of the mean. Student’s t-test was used to compare pre and post-infusion data, with p-values < 0.05 considered statistically significant.

III. RESULTS

Fig 2 illustrates how SOBI-MGG components reconstructed to the sensor array enable the identification of gastric slow wave components. Raw MGG signals from a normal subject, shown in Fig 2a, contain respiratory, cardiac, and gastric activity. SOBI components computed from the filtered MGG isolate specific signal components at different frequencies (Fig 2b; power spectra shown in Fig 2c). The reconstructed SOBI-MGG spatial maps and power spectra (Fig 2d and 2e) show gastric slow wave sources with a higher signal-to-noise ratio than raw data.

Figure 2:

(a) Raw MGG signals from 17 sensors in the SQUID magnetometer during baseline in a normal subject. (b) Seven SOBI components classified as gastric with (c) associated power spectra. (d) Reconstructed SOBI-MGG with (e) power spectra. The reconstructed SOBI-MGG is largely free of interference from respiratory, cardiac, and other noise contributions to the signals.

Representative data obtained from one of the normal human subjects before and after intravenous administration of erythromycin (3 mg/kg) is shown in Fig 3. Fig 3a-d shows data from the reconstructed SOBI-MGG/SOBI-EGG central channel and the corresponding power spectra before and after (t = 5, 20, 30 min) erythromycin administration. Filtered mucosal EMG signals with their corresponding FFTs at the same respective times are also presented in Fig 3e-f. Normal gastric slow wave activity near 3 cpm was clearly evident in all baseline recordings. After erythromycin was administered, MGG, EGG and EMG showed abnormal gastric activity that eventually reverted towards normal activity over time.

Figure 3:

(a-d) Representative SOBI-MGG/ SOBI-EGG signals and corresponding power spectra at successive time intervals before and after erythromycin (3mg/kg) administration. (e-f) Mucosal EMG signals with corresponding power spectra before and after erythromycin administration are shown for comparison. Abnormal gastric activity evident immediately after administration of erythromycin eventually returned towards normal activity 30 minutes post-administration in MGG, EGG, and EMG.

For 3 mg/kg erythromycin administration, slow wave frequencies computed from both MGG and EGG changed significantly post-erythromycin, as shown in Table 1. Similar changes were not observed in EMG slow wave frequencies. For 6 mg/kg erythromycin administration, EMG slow wave frequencies showed a significant decrease at t =1-5 min after infusion that persisted for 31-40 min. On the other hand, MGG slow wave frequencies showed statistically significant decrease only at 11-20 min after administration of 6 mg/kg erythromycin. A statistically significant decrease in EGG slow wave frequency was observed at 21-30 min after administration of 6 mg/kg erythromycin.

TABLE I.

Changes in waveform correlation, frequency and amplitude in EMG. EGG and MGG following erythromycin infusion.

Time/dosage	EMG/MGG WAVEFORM CORRELATION	EMG/EGG WAVEFORM CORRELATION	FREQUENCY (CPM)			AMPLITUDE
			EMG	EGG	MGG	EMG (mV)	EGG (mV)	MGG (pT)
Baseline
3 mg/kg	0.56±0.03	0.59±0.03	3.0±0.09	3.0±0.1	3.1±0.1	6.0±0.8	0.32±0.04	10.9±2.1
6 mg/kg	0.58±0.06	0.52±0.03	3.1±0.03	2.7±0.2	2.8±0.1	6.5±1.9	0.36±0.06	28.3±12.4
t=1-5 min
3 mg/kg	0.49±0.03 (p = 0.19)	0.52±0.02 (p = 0.06)	2.9±0.1 (p = 0.44)	2.4±0.2 (p < 0.01)	2.1±0.2 (p < 0.001)	13.9±2.4 (p < 0.01)	0.65±0.09 (p < 0.01)	9.7±1.3 (p = 0.64)
6 mg/kg	0.48±0.06 (p = 0.29)	0.47±0.05 (p = 0.46)	2.6±0.2 (p < 0.05)	2.5±0.5 (p = 0.77)	2.4±0.2 (p = 0.16)	11.3±2.6 (p = 0.16)	0.27±0.03 (p = 0.19)	13.4±2.6 (p = 0.33)
t=11-20 min
3 mg/kg	0.54±0.03 (p = 0.79)	0.49±0.05 (p = 0.14)	2.6±0.2 (p = 0.09)	2.5±0.1 (p < 0.05)	2.4±0.2 (p < 0.01)	14.3±2.9 (p < 0.05)	0.66±0.10 (p < 0.01)	9.5±1.6 (p = 0.61)
6 mg/kg	0.45±0.05 (p = 0.19)	0.54±0.13 (p = 0.90)	2.6±0.2 (p < 0.05)	2.3±0.3 (p = 0.34)	2.1±0.2 (p < 0.05)	16.3±2.9 (p < 0.05)	0.31±0.03 (p = 0.50)	14.3±2.5 (p = 0.35)
t=21-30 min
3 mg/kg	0.55±0.05 (p = 0.93)	0.51±0.07 (p = 0.45)	2.6±0.3 (p = 0.20)	2.2±0.2 (p < 0.01)	2.5±0.2 (p < 0.05)	11.2±0.9 (p < 0.001)	0.70±0.20 (p = 0.15)	9.6±1.3 (p = 0.60)
6 mg/kg	0.49±0.04 (p = 0.27)	0.45±0.03 (p = 0.20)	2.8±0.1 (p < 0.05)	2.0±0.2 (p < 0.05)	2.3±0.2 (p < 0.08)	13.8±1.8 (p < 0.05)	0.32±0.01 (p = 0.55)	13.6±2.6 (p = 0.33)
t=31-40 min
3 mg/kg	0.53±0.05 (p = 0.65)	0.54±0.08 (p = 0.54)	2.8±0.1 (p = 0.11)	2.1±0.3 (p = 0.05)	2.7±0.2 (p = 0.05)	7.2±0.9 (p = 0.33)	0.52±0.20 (p = 0.39)	6.0±1.0 (p = 0.07)
6 mg/kg	0.45±0.01 (p = 0.13)	0.43±0.03 (p = 0.11)	2.7±0.1 (p < 0.05)	2.5±0.4 (p = 0.73)	2.5±0.1 (p = 0.08)	8.8±1.1 (p = 0.35)	0.31±0.03 (p = 0.48)	19.5±6.4 (p = 0.57)

Values are means ± SEM. p values measure difference from baseline values.

The percent power distribution (PPD) in frequency ranges associated with normal slow waves, bradygastria, and tachygastria may indicate functional changes in the gastric slow wave [23]. In both MGG and EMG, the percentage of power in brady- and tachygastric regions tended to increase and power in regions associated with normal slow wave decreased after 3 mg/kg erythromycin infusion (see Fig 4 and Table 2). EGG analysis showed significant differences in brady- and normogastric PPD at t = 1-5 min, t = 11-20 min and t = 21-30 min post-erythromycin. On the other hand, EGG PPD analysis failed to demonstrate any significant differences in tachygastric PPDs during the course of the experiment.

Figure 4:

Percent power distributed (PPD) in brady-, normo-, and tachygastric frequency ranges for EMG, MGG and EGG recordings in normal human subjects at successive times both before and after erythromycin administration (3 mg/kg and 6 mg/kg). Statistically significant post-erythromycin PPD changes were denoted by *.

TABLE II.

Changes in PPD of EMG, MGG and EGG following erythromycin infusion

Time/dosage	PPD (%),f< 2 cpm			PPD (%),f=2-4 cpm			PPD (%),f> 4 cpm
	EMG	MGG	EGG	EMG	MGG	EGG	EMG	MGG	EGG
Baseline
3 mg/kg	6.8±1.2	10.4±1.7	13.0±2.1	86.6±1.7	78.5±1.7	69.4±2.8	6.7±1.1	11.1±2.3	17.6±2.3
6 mg/kg	6.3±1.5	5.7±2.0	14.3±3.4	83.0±3.6	81.4±2.9	71.5±3.9	10.8±2.9	12.9±2.1	14.2±1.9
t=1-5 min
3 mg/kg	16.9±2.7 (p < 0.01)	28.2±3.6 (p < 0.001)	27.9±4.5 (p < 0.01)	70.2±3.4 (p < 0.001)	51.5±4.0 (p < 0.001)	51.6±4.3 (p < 0.01)	12.9±1.6 (p < 0.01)	20.2±1.8 (p < 0.01)	20.5±2.3 (p = 0.38)
6 mg/kg	19.3±4.4 (p < 0.05)	20.4±6.4 (p = 0.09)	19.2±10.6 (p = 0.68)	67.1±4.4 (p < 0.05)	53.9±3.1 (p < 0.001)	52.9±8.8 (p = 0.13)	13.6±2.6 (p = 0.47)	25.7±4.2 (p = 0.06)	27.9±2.5 (p < 0.01)
t=11-20 min
3 mg/kg	20.6±3.3 (p < 0.001)	24.9±3.6 (p < 0.01)	23.3±2.3 (p < 0.01)	65.7±4.7 (p < 0.001)	53.8±3.5 (p < 0.001)	60.6±2.4 (p < 0.05)	13.7±2.3 (p < 0.05)	21.2±2.8 (p < 0.05)	16.0±0.9 (p = 0.55)
6 mg/kg	24.9±3.6 (p < 0.001)	28.7±6.2 (p < 0.05)	25.8±6.9 (p = 0.17)	55.6±4.0 (p < 0.001)	52.0±5.8 (p < 0.01)	50.9±5.4 (p < 0.01)	19.5±3.4 (p = 0.06)	19.3±4.3 (p = 0.22)	23.3±4.3 (p = 0.09)
t=21-30 min
3 mg/kg	24.3±5.1 (p < 0.05)	21.5±3.9 (p < 0.05)	29.0±5.1 (p < 0.05)	62.4±5.8 (p < 0.01)	58.5±4.2 (p < 0.001)	54.0±5.1 (p < 0.05)	13.3±1.5 (p < 0.01)	19.9±2.1 (p < 0.05)	17.1±1.9 (p = 0.87)
6 mg/kg	18.7±2.7 (p < 0.001)	29.0±3.1 (p < 0.001)	28.9±3.6 (p < 0.05)	61.5±2.8 (p < 0.001)	52.8±3.0 (p < 0.001)	51.8±2.2 (p < 0.01)	19.8±2.5 (p < 0.05)	18.2±2.7 (p = 0.15)	19.4±2.5 (p = 0.12)
t=31-40 min
3 mg/kg	14.7±2.7 (p <0.05)	15.7±2.9 (p = 0.15)	27.0±8.4 (p = 0.18)	72.6±5.8 (p = 0.06)	62.6±3.1 (p < 0.01)	55.7±8.0 (p = 0.17)	12.7±2.5 (p = 0.19)	21.7±2.5 (p < 0.05)	17.3±2.9 (p = 0.95)
6 mg/kg	17.4±3.6 (p < 0.05)	22.0±4.8 (p < 0.05)	31.2±4.1 (p < 0.01)	68.3±4.4 (p < 0.05)	60.6±2.6 (p < 0.01)	51.9±3.3 (p < 0.01)	14.3±1.6 (p = 0.3)	17.4±2.5 (p = 0.21)	16.9±1.5 (p = 0.27)

Values are means ± SEM. p values measure difference from baseline values.

For 6 mg/kg erythromycin infusion, brady- and normogastric PPDs computed from both MGG and EMG changed significantly post-erythromycin, as illustrated in Fig 4 and Table 2. However, normogastric EGG PPD shows significant changes starting at t= 11-20 min after infusion that persisted for 31- 40 min after infusion. Interestingly, there was no significant change in power at tachygastria frequencies after 6 mg/kg erythromycin infusion in MGG, EMG, or EGG recordings.

Pre-erythromycin MGG signal amplitudes did not change significantly after infusion of either 3 mg/kg or 6 mg/kg erythromycin. However, the EMG signal amplitude increased significantly immediately following the administration of 3 mg/kg erythromycin which reduced to normal values at 31- 40 min after infusion, as illustrated in Table 1. For 6 mg/kg erythromycin infusion, a statistically significant EMG amplitude increase was observed at t = 11-20 min post-erythromycin, but it reduced to normal values at 31- 40 min after infusion. We observed a statistically significant EGG amplitude increase after 3 mg/kg erythromycin infusion at t = 1-5 min that persisted for t = 11-20 min post-erythromycin. EGG amplitude returned towards normal values at t = 21-30 min (p=0.21) post-erythromycin. Interestingly, EGG amplitude changes were not observed to be significant following 6 mg/kg erythromycin infusion.

We also calculated the waveform correlations, i.e., the mean value of the maximum cross-correlation coefficient between EMG and SOBI-MGG/SOBI-EGG. Erythromycin did not change waveform correlation values between with EMG and SOBI-MGG or SOBI-EGG for either 3 mg/kg or 6 mg/kg dosages (see Table 1).

The use of SOBI-computed propagation maps from MGG data allowed us to visualize the propagating slow wave and compute the propagation velocity in normal subjects. In baseline data, SOBI MGG propagation maps reveal a left-to-right normal anterograde gastric propagation in four subjects. However, one volunteer presented a retrograde pattern. The average baseline propagation velocity, including the anomalous retrograde subject was 4.9 ± 1.7 mm/s. Propagation patterns and velocities at t = 11-20 min post infusion varied widely but on average were retrograde with an average velocity of −3.7 ± 2.5 mm/s and were significantly different from baseline values (p<0.05). We also observed static, rotational, and anterograde propagation patterns in some subjects. Subsequently, propagation velocity normalized at t=21-30 min after infusion to baseline values (p=0.66), and the predominant propagation pattern observed was anterograde. Fig 5 shows representative propagation patterns during pre and post erythromycin (3 mg/kg) infusion (baseline, t = 5min and t = 25 min) in a typical subject. Abrupt spatial discontinuities are evident in the temporal progression of MGG propagation patterns (fig 5a). The pattern appeared nearly static between 2 and 6 seconds, with maximum jumps pronounced between 6 and 8 seconds. It’s likely the limited hardware resolution imposed by the SQUID detector array combined with a potentially slower propagation at the beginning of the slow wave sequence prevents us from seeing subtle motion in the maps between 2 and 6 seconds.

Figure 5:

MGG propagation pattern maps during (a) baseline, (b) t = 5 min post-administration, and (c) t = 25 min post-administration of erythromycin (3 mg/kg) in a typical subject. Each row shows 5 maps of magnetic field activity at two-second intervals in a single subject. During baseline, anterograde propagation is evident as the pattern maxima moves towards the left side of the map, corresponding to the subject’s right side. Retrograde patterns were observed at t = 5 min post erythromycin administration. However at t = 25 min post-administration, the SOBI-MGG propagation pattern returns to normal anterograde propagation.

For the 6 mg/kg erythromycin infusion, baseline propagation patterns were anterograde with propagation velocities of 8.1 ± 0.6 mm/s. Retrograde propagation with propagation velocity −3.6 ± 2.9 mm/s (p < 0.05) was observed at t = 11-20 min after 6 mg/kg erythromycin infusion, similar to the propagation characteristics observed at t=11-20 min after 3 mg/kg erythromycin infusion. Unlike the 3 mg/kg erythromycin infusion, we also noted retrograde propagation with propagation velocity of −4.1 ± 2 mm/s (p < 0.01) at t = 21-30 min post infusion. Afterwards, propagation velocity normalized at t= 31-40 min to baseline values (p = 0.32) and propagation maps showed normal anterograde gastric propagation in most subjects.

IV. DISCUSSION

Functional alterations of dopaminergic (inhibitory) and muscarinic (stimulatory) receptors are thought to be implicated in the pathogenesis of gut dysmotility. Gastric dysmotility can be induced pharmacologically by agents such as epinephrine, glucagon, met-enkephalin and prostaglandin [24–26]. Erythromycin has a stimulating effect on gastrointestinal (GI) motility, acting as a motilin receptor agonist in the gut stimulating enteric nerves and smooth muscle and triggering a phase of the migrating myoelectric complex [27]. The use of erythromycin as a prokinetic agent for improvement of gastrointestinal motility disorders has been examined in a range of clinical settings [28–31]. Sarna et. al [32] investigated the gastrointestinal motor effects of erythromycin in healthy humans and reported that the degree and duration of the response of clinically effective doses of erythromycin in stimulating the antral motor activity is greater in the fasted than in the fed state. By contrast, it has a slight inhibitory effect on small bowel motor activity. In a study of acute and chronic treatment of gastroparesis with erythromycin, Richards et al. observed that acute intravenous dosing of erythromycin accelerated gastric emptying in most patients with gastroparesis [33].

Though several authors have investigated the effect of erythromycin on GI motility, few studies have reported the effect of erythromycin has on gastric slow wave activity or the cutaneous EGG [8], and no such studies have appeared for the MGG. The stomach generates rhythmic electrical impulses, known as slow waves that can coordinate mechanical contractions. Slow wave activity initiates along the greater curvature in the upper corpus region and propagates toward the pylorus with increasing velocity and amplitude [34–36]. Slow wave dysrhythmias are believed to contribute to dysmotility symptoms such as gastroparesis. Erythromycin seems to affect gastric slow wave activity differently in healthy subjects than in patients with gastroparesis; studies by Chen et al. observed irregularities in the cutaneous EGG which were more pronounced during the first hour of intravenous erythromycin administration (6mg/kg) in normal human subjects. [8]. However in patients with gastroparesis, infusion of erythromycin was able to normalize gastric dysrhythmias [37].

In this study, the effect of erythromycin on gastric slow wave activity on the non-invasive MGG and EGG was quantified and results were validated using mucosal EMG. EMG, MGG, and EGG reflected normal gastric slow wave activity during baseline recording in all normal subjects, and we saw abnormalities in the slow wave activity after intravenous administration of both 3 mg/kg and 6 mg/kg erythromycin. For 3 mg/kg erythromycin infusion, post-erythromycin MGG and EMG PPDs showed significantly less power in the normogastric frequency range and correspondingly more power in both brady and tachygastric ranges, respectively. We also noted PPD changes in brady- and normogastric EGG frequency ranges after 3 mg/kg erythromycin infusion; however, no significant difference in EGG tachygastric PPDs was observed. For 6 mg/kg erythromycin infusion, MGG, EMG, and EGG recordings showed bradygastric and normogastric PPD changes. However, unlike the MGG and EMG 3 mg/kg dosage, tachygastric frequencies did not show significant differences. In our studies we defined the range of gastric frequencies more broadly than other studies (1-9 cpm) to ensure capture of any tachygastnc signals, interestingly, we observed tachygastric PPD changes in 3 mg/kg post-erythromycin but not the 6 mg/kg dosage. Similar EGG studies by Chen et al showed significant differences in bradygastric and normogastric PPDs after intravenous erythromycin administration (6mg/kg) which they proposed to result from the electrical response activity associated with strong low frequency contractions induced by erythromycin. It is possible that contractile activity introduces motion artifact which may obscure some of the normal 2 - 4 cpm activity [8]. Previous animal studies also reported bradyarrhythmias with erythromycin infusion, attributed to retrograde giant contractions, with retelling and sometimes with vomiting [38].

We observed a statistically significant EMG amplitude increase immediately after infusion of both 3 mg/kg and 6 mg/kg erythromycin which reduced to normal values at 31- 40 min after infusion. However, no statistical difference was observed for MGG amplitude before or after administration of either 3 mg/kg or 6mg/kg erythromycin. It is possible that increased serosal EMG amplitudes are caused by pyloric contractions which do not significantly affect MGG signals. Chen et al. noted an amplitude increase of the EGG with intravenous 6mg/kg erythromycin infusion which was attributed to the increased contractile activity of the stomach [8]. We also observed a significant EGG amplitude increase immediately after 3 mg/kg erythromycin infusion which reverses towards normal values at 21-30 min after infusion. Surprisingly, we did not observe EGG amplitude changes following 6 mg/kg erythromycin infusion, which raises the question of potential variability in the effects of mechanical contractions on gastric slow wave amplitudes. This suggests more precise studies with combined electrical and mechanical measurements are needed.

The cross-correlation of signal waveforms shows that specific waveform features captured by mucosal EMG are present in both EGG and MGG. Further, the waveform correlation between EMG and SOBI MGG or SOBI EGG did not show any significant difference before and after administration of either 3 or 6 mg/kg erythromycin.

Since one subject consistently exhibited retrograde propagation during baseline, the average propagation velocity computed from the baseline data for 3 mg/kg study group was slightly lower than that for 6 mg/kg study group. However if the outlier is excluded, both study groups have average velocities (7.0 ± 1.2 mm/s, 8.1 ± 0.6 mm/s respectively) consistent with normal slow wave propagation velocities as reported previously using MGG [18]. In that study, propagation velocity was anterograde at 7.4 mm/s for control subjects. Using serosal recordings, O’Grady et al. observed a distinct gradient in propagation velocity; a mean slow-wave propagation velocity of 8.0 mm/s in the pacemaker region, decelerating to 3.0 mm/s in the corpus, and again increasing to 5.9 mm/s in the antrum for normal subjects [36]. Recently, a study using HR - EGG also found an average speed of 3.7 mm/s consistent with O’Grady’s findings [11]. The spatial resolution of our from our 17 channel SQUID detector array facilitated only a single global estimate of propagation velocity, but a larger density sensor array could provide an estimate of the velocity gradient. Future studies using high density MGG and EGG are required to address the discrepancy observed in these modalities. We observed a decrease in the propagation velocity of the gastric slow wave in SOBI-MGG after administration of erythromycin. Interestingly, these changes were not observed to be significant statistically until 11-20 min after administration of both 3 mg/kg and 6 mg/kg erythromycin. We observed several different types of propagation patterns in post-erythromycin while pre-erythromycin generally showed consistent left-right propagation. Propagation velocity increased towards baseline values after 21-30 min in 3mg/kg and after 31-40 min in 6 mg/kg erythromycin infusion.

Although we are always able to detect and record MGG, there are several technical limitations that have prevented MGG from becoming a widely used diagnostic tool for functional gastrointestinal diseases. SQUID magnetometers are currently expensive and require the use of magnetic shielding. The addition of a helium reliquifier, which captures and reliquifies the helium gas bum off can alleviates the high cost of liquid helium. Additionally, lower-cost alternatives for biomagnetic applications including atomic magnetometers and high-temperature SQUIDs are becoming available. High density MGG and EGG mapping integrated with SOBI signal processing techniques represents an opportunity to noninvasively assess the potential contribution of gastric slow wave dysrhythmias to the pathophysiology of functional gastric disorders.

IV. CONCLUSION

In conclusion, this study investigated the effect of erythromycin on gastric slow wave parameters using non-invasive MGG/EGG validated with mucosal EMG. These results encourage further investigation of the bio-electromagnetic effects of erythromycin on the gastric slow wave in gastroparesis patients. Functional spatiotemporal assessment of electrical activity in the stomach with positive predictive value could promote the diagnosis of gastroparesis from its current status as a diagnosis of exclusion.