Background
Preterm birth is the largest single cause of infant death in the United States. A cervical length of <2.5 cm, measured in the mid-trimester, has been shown to identify individuals at increased risk. Uterine electromyography is an emerging technology for noninvasively assessing uterine bioelectrical activity. With its ability to characterize nuanced differences in myometrial signals, uterine electromyography assessments during the mid-trimester may provide insight into the mechanisms of cervical shortening.
Objective
This study aimed to characterize uterine bioelectrical activity in pregnant individuals with short cervices in the mid-trimester compared with that of pregnant individuals of the same gestational age with normal cervical lengths.
Study Design
This is a prospective cohort study of subjects with singleton, nonanomalous pregnancies between 16 weeks and 0 days and 22 weeks and 6 days of gestational age. Subjects with normal cervical length (≥3.0 cm) were compared with subjects with short cervical length (<2.5 cm). The short-cervical-length cohort was further stratified by history of preterm birth. Multichannel uterine electromyography recordings were obtained for ∼60 minutes using proprietary, directional electromyography sensors on the abdomen. Uterine electromyography signals were observed and classified in groups as spikes, short bursts, and bursts. Primary outcomes were relative expression of spike, short-burst, and burst uterine electromyography signals. Subgroup analyses assessed each signal percentage by cervical length, history of preterm birth, and gestational age at delivery. Differences in percentage of uterine electromyography signals according to cervical length were analyzed using nonparametric tests of significance.
Results
Of the 28 included subjects, 10 had normal and 18 had short cervical length. There were 9 subjects with short cervical length and a history of preterm birth. Spikes were the most commonly recorded signals and were higher in the normal-cervical-length cohort (96.3% [interquartile range, 93.1%–100.0%]) than the short-cervical-length cohort (75.2% [interquartile range, 66.7%–92.0%], P =.001). In contrast, median percentages of short-bursts and bursts were significantly higher in subjects with a short cervical length (17.3% [interquartile range, 13.6%–23.9%] vs 2.5% for normal cervical length [interquartile range, 0%–5.5%], P =.001 and 6.6% [interquartile range, 0%–13.4%] vs 0% for normal cervical length [interquartile range, 0%–2.8%], P =.014, respectively). Within subgroup analyses, cervical length was inversely proportional to percentage of observed short-bursts ( P =.013) and bursts ( P =.014). Subjects with short cervical length and history of preterm birth had higher burst percentages (12.8% [interquartile range, 9.0%–15.7%]) than those with short cervical length and no history of preterm birth (3.3% [interquartile range, 0%–5.0%], P =.003).
Conclusion
Short-burst and burst uterine electromyography signals are observed more frequently in mid-trimester patients with short cervical lengths. This relationship provides insight into abnormal myometrial activation in the mid-trimester and offers a plausible biophysiological link to cervical shortening.
Introduction
Preterm birth (PTB) is the largest single cause of infant death in the United States , and the leading cause of neurodevelopmental disabilities in surviving infants. Although the definitive etiology of PTB remains elusive, cervical length (CL) shortening has been shown to identify those at increased risk. Among all pregnant people, 18% with cervical shortening <2.5 cm and 50% with a CL of <1.3 cm in the mid-trimester (16 weeks 0 days to 22 weeks 6 days gestational age [GA]) will deliver prematurely if not treated. Mechanisms of pathologic cervical shortening in the mid-trimester are, however, thus far unclear.
Why was this study conducted?
To characterize uterine bioelectrical activity in mid-trimester pregnancies with short cervical lengths using directional electromyographic sensors.
Key findings
Short-burst and burst uterine electromyography signals are observed in patients with short cervical lengths. This relationship suggests that abnormal myometrial activation begins in the mid-trimester and is either caused by or potentially causes cervical shortening.
What does this add to what is known?
Uterine electrophysiological recordings can identify myometrial activity in the mid-trimester. Our data enhance understanding of tissue-level myometrial electrophysiology and offer a plausible biophysiological link to mid-trimester cervical shortening.
At term, the onset of labor is associated with increased expression of uterine bioelectrical activity, which can be recorded by electromyography (EMG). These uterine EMG (uEMG) signals can be characterized as spikes, Short-bursts, or bursts, which can be distinguished using well-defined criteria ( Figure 1 ). , The spike signal is an isolated voltage transient that lasts between 3 and 15 seconds and has sharp upward and downward phases; it is representative of a single action potential. Short-bursts are composed of 2 or more spikes that, in total, last ≤20 seconds. Bursts are repetitive spikes, lasting >20 seconds; these signals likely represent tissue-level contractions. The progression from spikes to bursts likely correlates with increasing numbers of cells or larger areas of myometrial activation. ,
However, limited data describing uEMG signals recorded in the mid-trimester are available. Grgic and colleagues , have authored the 2 studies assessing mid-trimester uEMG activity. Although they suggest that pregnant individuals with shorter CL have greater uEMG activity, their work is limited to the recording of only spike signals and does not assess the relationship between presence of signals and the pathologic short cervix.
The purpose of this study is to compare the relative expression of spikes, short-bursts, and bursts in mid-trimester pregnancies with a short CL (<2.5 cm) with the signal expression in individuals of the same GA range with a normal CL. We hypothesize that the relative expression of each uEMG signal type will be different in individuals with short CL compared with those with normal CL.
Materials and Methods
This cohort study was approved by our institutional Research Subjects Review Board, and all subjects provided written informed consent for study participation. This study is reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
We performed a prospective cohort study of subjects with singleton, nonanomalous pregnancies and GA between 16 weeks and 0 days and 22 weeks and 6 days receiving care at an urban academic medical center between November 2018 and March 2021. Participants were recruited into the following cohorts: (1) normal CL (≥3 cm) and (2) short CL (<2.5 cm). Subjects in the short-CL cohort were further stratified into subjects with and without history of spontaneous PTB; Appendix A contains the details of obstetrical history. Each enrollment group (normal CL, short CL with no PTB history, and short CL with PTB history) had a recruitment goal of 10 subjects to be able to assess the impact of history of PTB. PTB was defined as any delivery at GA <37 weeks.
At our institution, CL screening via transvaginal ultrasound is offered under 2 circumstances: (1) universal CL screening at the time of the anatomic survey and (2) serial CL assessments for patients with history of PTB, which begin at 16 weeks of GA. Transvaginal sonography was performed with the GE Voluson E10 RIC5-9D 3D endovaginal probe (GE Healthcare, Milwaukee, WI). Image criteria for CL measurement were consistent with the Cervical Length Education and Review (CLEAR) guidelines. At least 3 measurements were made and the shortest, best measurement was reported. Short CL was defined as length <2.5 cm (≤5th percentile for our population). All CL measurements were reviewed by University of Rochester Maternal-Fetal Medicine Faculty. In addition, a random sampling of CL measurements was reviewed by 1 of the authors (P.M.) and found to be in agreement with initial reports.
Eligible individuals with normal or short CL were approached as they were identified and were assigned into the above groups. All eligible subjects were required to be English-speaking and reading and aged ≥18 years. Subjects with known uterine anomalies, a history of cervical surgery, including loop electrosurgical excision procedure or cervical conization, and body mass index (BMI) >38.0 kg/m 2 were excluded from the study. Patients with higher BMI were excluded to limit variation in distance from skin to uterine surface; unpublished experience from our previous studies has shown that uEMG sensor performance is not influenced by habitus up to a BMI of 38.0 kg/m 2 .
Subjects using medications that may act as a uterine tocolytic (ie, calcium channel blockers, magnesium supplements, cyclooxygenase inhibitors, beta-agonists) and those that had already begun treatment with vaginal progesterone were also excluded.
Notably, individuals maintained on 17-hydroxyprogesterone (17-OHP) remained eligible because the use of this agent was not considered an intervention for short CL but was rather used as prophylaxis for prevention of recurrent PTB. Subjects were also excluded if they demonstrated imminent risk of delivery (ie, contractions, bleeding, loss of fluid, etc.).
Our electronic medical record was used to corroborate and obtain other required variables, including race, ethnicity, age, height, weight, gravidity and parity, GA, and obstetrical and medical histories.
Pregnancy outcomes for each subject were obtained, including use of interventions (vaginal progesterone and/or cerclage), GA at delivery, obstetrical outcomes, and pregnancy complications. For subjects with unavailable outcome data through electronic medical records, we attempted to contact them by telephone with a designated script approved by our Research Subjects Review Board.
Uterine electromyography assessment
Electromyographic sensors comprised a proprietary, 8-cm-diameter uEMG “Area Sensor” paired with a commercially available echocardiogram (ECG) pad. Area Sensors are constructed of the same material as standard EMG sensors but have a central opening; they were approved for human use at the study institution. The ECG pad was placed within the open space at the center of the Area Sensor. When the signals from the Area Sensor were referenced to the central ECG pad, the resulting differential signal (also called the “Laplacian” signal) preferentially reported the bioelectrical signals produced by the portion of the uterine wall located immediately beneath the sensors. Each uEMG Area Sensor–ECG pad pair, therefore, reported uterine signals that originated from different locations of the uterine wall. In the first 2 subjects, we attempted to record uEMG signals from 4 sensor pairs placed over the small mid-trimester uterus. However, on the basis of recording quality, it became apparent that there was inadequate space to accommodate 4 Area Sensors and simultaneously record contractions with a tocodynamometer (toco). We subsequently used 2 Area Sensor —ECG pad pairs for each of our remaining subjects ( Appendix B ).
Recordings were obtained for a minimum of 60 minutes using the TMSi Porti EMG amplifier (TMSi, Oldenzaal, Netherlands), which is approved by the United States Food and Drug Administration for human EMG recordings. Subjects were monitored in a reclining position throughout the duration of the session. All uEMG data were recorded from direct current to >1000 Hz, then digitally filtered between 0.15 and 1.2 Hz to isolate uterine signals for analysis. Frequencies selected for analysis at viable gestations are commonly performed at 0.3 to 1.0 Hz; our band-pass was expanded to account for the possibility that important mid-trimester signals may include a broader range of frequencies.
Software was provided by TMSi Polybench software (TMSi, Oldenzaal, Netherlands). Root mean square processing techniques were applied to all data configurations to provide a measure of power for the signals.
To assess uterine activity, we first defined each signal type on the basis of characteristics previously defined in term labor ( Figure 1 ). , Signals were considered to be present if the signal-to-noise ratio was >2.5 and the voltage amplitude was >15 μV (baseline to peak); 15 μV was identified as the signal threshold that eliminated the small spurious spikes commonly seen in background noise. Because burst voltages rarely exceeded 250 μV, voltage transients >500 μV were considered artifacts. Each signal type was identified by visual inspection of uEMG signals by 1 of the authors. When classification was not clear, a second expert opinion was obtained. Signals occurring from both uEMG Area Sensor–ECG pad pairs were analyzed simultaneously. To avoid double counting, signals appearing within 1 second of each other were considered to be the same signal, even if coming from different Area Sensor–ECG pad pairs.
The primary outcome was relative expression (as a percentage of total signals) of spikes, short-bursts, and bursts. Subgroup analyses assessed signal percentage by CL, history of PTB, and GA at delivery.
In the setting of limited previous data for estimating size or interquartile range, sample size was determined using Mead’s Resource Equation. The total recruitment goal was 30 subjects and 3 groups (normal CL; short CL with no PTB history; short CL with PTB history), with 10 subjects in each group. Mead’s Resource Equation, thus, yields an E of 27, which indicates the presence of a sufficient sample size (goal E=10–20).
Univariate analyses were performed for demographic data with P <.05 considered significant. For demographic data, Student’s t test was used for continuous comparisons and Fisher exact test for categorical comparisons.
Each signal type was reported relative to the total quantity of signals for individual subjects. Uterine EMG signal distributions across our study population were displayed as histograms to determine normality of distribution. Given that our uEMG data were nonparametric, we used Mann–Whitney U and Kruskal–Wallis tests to compare median percentages of signal types. CL was analyzed as a continuous variable using linear regression. Additional pairwise comparisons and regression analyses were performed to assess the influence of BMI, GA, and 17-OHP on signal expressivity.
Results
A total of 28 subjects, 10 in the CL ≥3 cm cohort and 18 in the CL <2.5 cm cohort, provided consent and completed our study with at least 60 minutes of uEMG data recording. A flowchart depicting subject selection is shown in Figure 2 .
Detailed maternal demographic characteristics are summarized in Table 1 ; baseline data of the normal and short-CL groups are overall similar. Pregnancy outcomes were also comparable between the 2 cohorts ( Table 2 ). In the short-CL cohort, 9 of 18 subjects had a history of PTB, and 4 of these were prescribed intramuscular 17-OHP for risk reduction. The relatively low rate of 17-OHP use was due to a change in recommendations at our institution based on data published in the 17-OHCP to Prevent Recurrent Preterm Birth in Singleton Gestations (PROLONG) study. Patients who were candidates for prophylaxis early in the study were not routinely offered prophylaxis late in the study. Average CL for the control group was 3.9 ± 0.5 cm vs 1.8 ± 0.5 cm in the short-CL group. When stratified by history of PTB, the short-CL cohort with no history of PTB had an average CL of 2.0±0.3 cm. Comparatively, the short-CL cohort with history of PTB had a statistically significant lower average CL of 1.6±0.5 cm ( P =.074) ( Table 3 ).
| Maternal demographic characteristics | Cervical length ≥3.0 cm (n=10) | Cervical length <2.5 cm No history of PTB (n=9) | Cervical length <2.5 cm History of PTB (n=9) | P value |
|---|---|---|---|---|
| Age (y) | 30±5 | 28±7 | 29±6 | .58 |
| Race | ||||
| White | 10 (100.0) | 6 (66.7) | 5 (55.6) | .06 |
| Black | 0 (0) | 3 (33.3) | 3 (33.3) | |
| Asian | 0 (0) | 0 (0) | 1 (11.1) | |
| Non-Hispanic | 10 (100.0) | 9 (100.0) | 9 (100.0) | 1.0 |
| Body mass index (kg/m 2 ) | ||||
| <30.0 | 7 (70.0) | 5 (55.6) | 4 (44.4) | .71 |
| 30.0–34.9 | 1 (10.0) | 3 (33.3) | 3 (33.3) | |
| 35.0–38.0 | 2 (20.0) | 1 (11.1) | 2 (22.2) | |
| Parous | 5 (50.0) | 4 (44.4) | 9 (100.0) | .02 |
| Gestational age at study recording | 19.9±1.1 | 21.0±1.0 | 19.4±1.8 | .06 |
| Preexisting diabetes mellitus | 0 (0) | 0 (0) | 1 (11.1) | 1.0 |
| Preexisting hypertension | 0 (0) | 0 (0) | 1 (11.1) | 1.0 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
