The opioid epidemic continues to have a profound effect on the U.S. population, with more than 100 people dying of opioid overdose each day [
1,
2]. Infants experience secondary consequences of opioid use by adults. For example, 14–22% of women fill an opioid prescription during pregnancy, and one infant with prenatal exposure to opioids is born every 15 minutes [
3–
6]. Children with exposure to opioids in utero show lower school achievement and higher rates of behavioral problems in comparison with children without such exposure [
7,
8]. The exact mechanisms underlying these differences in outcomes are not well understood. However, infants with exposure to opioids in utero have a lower birth weight and a smaller head circumference at birth than infants without exposure to opioids in utero [
9,
10]. Moreover, at 4–8 weeks old, infants with opioid exposure, compared with infants without opioid exposure, have smaller regional brain volumes in multiple areas (despite no difference in overall brain volume), increased white matter injury, and altered functional networks on resting-state functional MRI [
11–
13]. Although opioid exposure has shown deleterious effects on health and development in early childhood, little is known about potential corresponding prenatal antecedents and whether or not the abnormalities observed in infants are already present in utero.
Fetal MRI is a powerful tool for examining the fetal brain, providing biometric measurements and assessment of brain morphology in vivo. The use of fetal MRI has been shown to result in improved diagnostic accuracy and confidence in detecting brain abnormalities before birth [
14]. Robust literature also supports the use of fetal MRI for determining normative 2D measurements of the brain [
15–
17]. Such measurements could be applied for the evaluation of fetuses with opioid exposure in utero (hereafter opioid-exposed fetuses).
The purpose of the present study was to compare opioid-exposed fetuses with fetuses without opioid exposure in utero (hereafter unexposed fetuses) in terms of 2D biometric measurements of the brain and additional pregnancy-related assessments on fetal MRI.
Methods
Study Design and Patients
This prospective case-control multiinstitutional study was HIPAA compliant and was approved by the institutional review board at each institution. Written informed consent was obtained from all study participants. Participants were recruited from three U.S. academic medical centers: Cincinnati Children's Hospital Medical Center, the University of Arkansas for Medical Sciences (hereafter referred to as Arkansas Children's Hospital), and the University of North Carolina at Chapel Hill, from July 1, 2020, through December 31, 2021. Patients in the third trimester of pregnancy were recruited to undergo fetal MRI for investigational purposes. Recruited patients were assigned to one of two study groups: those with opioid use during pregnancy and a control group without opioid use during pregnancy. To recruit patients with opioid use during pregnancy, flyers seeking volunteers were posted in obstetrics clinics and at substance use treatment programs. To recruit patients without opioid use during pregnancy, flyers seeking volunteers were posted in obstetrics clinics, e-mails seeking volunteers were sent to university employees, and previous research participants who agreed to be contacted for future research studies were contacted.
The recruitment materials indicated the following initial eligibility criteria: age of at least 18 years old, singleton pregnancy, and gestational age (GA) of at least 26 weeks. GA was self-reported by each potential participant and was confirmed in the electronic medical record (EMR) at each site by a nonauthor study coordinator, on the basis of the best obstetric estimate made using the first day of the last menstrual period or the earliest performed fetal ultrasound examination. The study coordinator conducted a telephone interview with potential participants before scheduling investigational fetal MRI, to explain the study and to assess the following additional screening criteria: inability to supply the name of at least one additional person to contact in the event that the participant could not be reached, known genetic disorder in the potential participant, fetal abnormality identified on prenatal ultrasound, nonviable fetus, contraindication to MRI, and inability of the participant to enter the magnet bore due to body habitus. Individuals were considered ineligible if they met any of these criteria. During this telephone interview, the study coordinator also informed potential participants that they would undergo a further interview regarding opioid exposure at the time of the fetal MRI appointment. The total number of potential participants who were screened for eligibility by telephone interview at each site, including the number of potential participants found to be ineligible as well as the number of potential participants who were eligible but declined to participate further, was not tracked.
At the start of the appointment, patients signed written informed consent to undergo fetal MRI. At the time of the informed consent discussion, before fetal MRI was performed, the study coordinator at each site also asked patients whether they used opioids during the pregnancy. Patients who reported opioid use were further questioned regarding the type of opioid used. Specifically, the patient was asked whether they used each of the following: codeine, heroin or morphine, fentanyl, or oxycodone. All patients, regardless of opioid use status, were also asked about their use of the following during pregnancy: nicotine, alcohol, marijuana, cocaine, amphetamine or methamphetamine, barbiturates, and benzodiazepines. All questions, other than the initial question regarding use of opioids during pregnancy, were optional. In addition, patients who reported opioid use were not required to report at least one specific type of opioid used. The question regarding alcohol exposure was only posed to patients enrolled at Cincinnati Children's Hospital Medical Center and was introduced at this center during the course of the study. The study coordinator also asked patients about the presence of gestational diabetes.
At the conclusion of the study, the study coordinator at each site also recorded birth weight when it was available on the basis of birth documents in the EMR. Newborns were classified as having fetal growth restriction if they had a birth weight below the third percentile based on the Fenton growth curve [
18].
MRI Acquisition
Fetal MRI examinations were performed using a 3-T system (Ingenia, Philips Healthcare) at Cincinnati Children's Hospital and a 3-T system (Prisma, Siemens Healthcare) at Arkansas Children's Hospital and the University of North Carolina at Chapel Hill. All three sites used a phased-array abdominal imaging coil. Patients were not sedated during the examinations. Patients were placed in the left lateral decubitus position, unless they reported feeling more comfortable in the supine position.
Examinations included localizer sequences in three orthogonal planes angled to the uterus, followed by a sagittal SSFP sequence of the uterus with 5-mm interleaved contiguous slices. SSFSE sequences of the fetal brain were obtained in the axial, sagittal, and coronal planes with 2-mm interleaved contiguous slices (
Figs. 1 and
2). Acquisition parameters were standardized across the three participating sites. At Cincinnati Children's Hospital, a radiologist assessed image quality in real time during the examinations; sequences were repeated until the monitoring radiologist was satisfied with image quality, with attention given to the midline sagittal image of the brain. At Arkansas Children's Hospital and the University of North Carolina at Chapel Hill, images were not assessed in real time by a radiologist.
MRI Interpretation
The results of the fetal MRI examinations were used for investigational purposes only. The images from all sites were submitted to a central site (Cincinnati Children's Hospital) for review. At the central site, two board-certified radiologists (U.D.N. and B.M.K.-F., with 8 and 18 years of posttraining experience), both of whom had added qualifications in pediatric radiology and fellowship training in pediatric neuroradiology, independently reviewed the examinations in an anonymized fashion by use of a research PACS. Before the start of image interpretation, the two radiologists discussed a standardized approach to the study measures. The radiologists were blinded to the clinical data and opioid use status of the patients. Aside from the assessment of interrater reliability, all study analyses reflected the interpretations of one investigator (U.D.N). That investigator (U.D.N.) repeated the measurements after an interval of approximately 1 month, to assess intrarater reliability.
Thirteen biometric parameters of the brain were measured manually using measurement tools in the PACS and were recorded (all expressed in millimeters): bone frontooccipital diameter (FOD), cerebral FOD, bone biparietal diameter (BPD), brain BPD, right cerebral BPD, left cerebral BPD, corpus callosum length, vermis height, anteroposterior (AP) vermis measurement, AP pons measurement, transverse cerebellar diameter, right atrial diameter, and left atrial diameter. Using the 2D measurements, four indexes were calculated to assess the relationship between the intracranial CSF spaces and the size of the supratentorial brain: frontooccipital index ([bone FOD − cerebral FOD] / bone FOD), biparietal index ([bone BPD − cerebral BPD] / bone BPD), right atrial index (right atrial diameter / right cerebral BPD), and left atrial index (left atrial diameter / left cerebral BPD). Previous investigators have described the methods used for determining these measurements and indexes [
15,
19,
20]. The midline area of the vermis was manually traced using external software (Intel-liSpace Portal, version 10.1, Philips Healthcare) and recorded in square millimeters.
The radiologists assessed additional pregnancy-related findings on MRI, including the degree of fetal motion, fetal positioning, cervical length, and amniotic fluid volume. The degree of fetal motion was subjectively assessed as decreased (little to no appreciable motion artifact), increased (motion artifact substantially compromising image quality), or normal (not meeting criteria for a decrease or increase). Fetal positioning was classified as cephalic or breech. Cervical length was measured in centimeters on a sagittal image of the uterus. Amniotic fluid volume was subjectively assessed as decreased (contact with the uterine wall by most of the fetal surface area), increased (as much or more amniotic fluid than the fluid volume of the fetus), or normal (not meeting the criteria for a decrease or increase). In addition to the subjective assessment of amniotic fluid volume, a deepest vertical pocket of amniotic fluid was measured on sagittal images of the uterus and recorded in centimeters.
One of the two investigators (U.D.N.) assessed fetal sex on the basis of MRI findings. If fetal sex could not be determined by MRI, then the study coordinator from the site determined this information by reviewing the EMR.
Statistical Analysis
Continuous variables were presented as mean ± SD, and categoric variables were presented as number and percentage. A two-sample
t test was used to compare continuous variables between patients with and without opioid exposure during pregnancy. Categoric variables were compared between the two groups using the chi-square test or the Fisher exact test. Given low numbers, all substances other than opioids and nicotine were grouped as other substances for purposes of comparison between the groups. Multivariable linear regression models were used to assess for significant differences between the groups for each brain biometry measurement and index (both treated as continuous variables), adjusting for GA (continuous variable), fetal sex (binary variable), and nicotine exposure (binary variable). Interrater and intrarater reliability were evaluated for continuous MRI measurements using intraclass correlation coefficients (ICCs) and for categoric assessments using percentage concordance, kappa coefficients, and weighted kappa coefficients. Agreement for continuous measurements was categorized on the basis of ICC as follows [
21]: poor, less than 0.40; fair, 0.40–0.59; good, 0.60–0.74; and excellent, 0.75 or greater. A
p value of less than .05 was considered statistically significant. Agreement was categorized for categoric measurements on the basis of percentage concordance and was considered excellent when greater than 75% [
22]; the kappa coefficients were not used for categorizing agreement of categoric measurements [
23]. All analyses were performed using SAS (version 9.4, SAS Institute).
Discussion
In the present study, we compared biometric measurements of the brain on fetal MRI, performed during the third trimester, between fetuses with and without in utero opioid exposure. The opioid-exposed fetuses, compared with unexposed fetuses, showed significantly smaller values for multiple brain biometry measurements as well as a significantly higher frontooccipital index. The opioid-exposed fetuses also showed a significantly higher frequency of breech presentation and a significantly higher frequency of subjectively increased amniotic fluid volume. The findings provide insight into the impact of prenatal opioid exposure on fetal development.
During the ongoing opioid epidemic in the United States, a growing body of literature has explored brain abnormalities on MRI in infants with prenatal opioid exposure. For example, studies have shown an increased incidence of white matter injury in infants with prenatal opioid exposure compared with unex-posed control infants, decreased neonatal head circumference in infants with prenatal opioid exposure, and significantly smaller head circumference in infants with prenatal opioid exposure and withdrawal symptoms severe enough to require medical management [
10,
12,
24–
26]. Additional studies described decreased regional brain volumes in neonates and school-age children with prenatal opioid exposure, particularly those involving the deep gray structures [
11,
27]. In another study, arterial spin-labeling showed increased global cerebral blood flow in opioid-exposed infants compared with control infants [
23]. Finally, multiple studies have shown alterations in functional connectivity on resting-state functional MRI in opioid-exposed neonates compared with control neonates [
13,
28–
30].
Although these prior studies have provided important advances, a full understanding of how opioids affect the developing brain also requires prenatal studies that evaluate the brain before the effects of opioid withdrawal and other postnatal variables (e.g., type of infant feeding, neonatal ICU stay, and maternal-infant bonding) can be assessed. Fetal MRI is the optimal imaging modality for studying the developing brain in vivo. Despite the current widespread availability, overall ease of performance, and safety profile of fetal MRI, fetal MRI studies of the brain of opioid-exposed fetuses are scarce. To our knowledge, the current study represents one of the largest such studies to date and adds to the prior literature by illustrating that impaired head growth begins in utero. A pilot study compared some of the measurements assessed in the present study between 12 opioid-exposed fetuses and 16 unexposed fetuses, observing a smaller AP vermis measurement in opioid-exposed fetuses [
31]. In comparison, the current study observed a significant difference in the vermis height between groups but not in the AP vermis measurement. The observation of a relatively small head size in utero may relate in part to abnormalities during neuronal proliferation [
32]. Indeed, evidence of impaired neurogenesis, proinflammatory changes, and increased programmed cell death (apoptosis) has been shown in rats with prenatal opioid exposure [
33–
36].
The cause of the increased frequency of breech presentation in opioid-exposed fetuses compared with unexposed fetuses is unclear. Breech presentation has been shown to be associated with a higher risk of congenital anomalies as well as obstetric risk factors such as oligohydramnios, fetal growth restriction, and gestational diabetes [
37–
39]. Though the available literature does not show a direct association between isolated prenatal opioid exposure and breech presentation, increased incidence of breech presentation has been described with illicit drug use and caffeine consumption during pregnancy [
40,
41]. Opioid exposure during pregnancy may cause a disruption in fetal neuromuscular development, potentially at the microstructural level, that results in breech presentation [
42,
43]. In addition, in rat models, it has been shown that alterations in brain myelination in opioid-exposed fetuses may lead to disrupted or dysfunctional movement [
44]. Although dolichocephalic molding is typical in fetuses with breech positioning, breech presentation seems unlikely to explain the higher frontooccipital index in opioid-exposed fetuses given that bone FOD was not significantly different between the two groups. Despite the greater frequency of breech presentation in opioid-exposed fetuses, the clinical relevance of this difference is uncertain given that the mean GA for the study sample was 32.0 weeks.
The cause of the higher frequency of increased amniotic fluid in opioid-exposed fetuses is also uncertain and may relate to decreased fetal swallowing secondary to cerebral dysfunction. Fetal amniotic fluid volume is also a function of fetal urine production and transplacental fluid balance; the potential impact of opioids on these mechanisms is unknown. In the present study, amniotic fluid volume was assessed using only qualitative and semiquantitative methods. However, these approaches are generally accepted in clinical practice given the absence of widely available reproducible quantitative methods [
45,
46].
The present study had limitations. First, the primary limitation was the small sample size, particularly in comparison with the size of prior studies performed in postnatal patient samples, despite the use of a multiinstitutional design. Future studies with larger samples of fetuses are needed to validate the findings. Second, we did not assess the impact of the length of in utero exposure, the GA when the exposure occurred, or the specific type of opioid to which the fetus was exposed. Third, among opioid-exposed fetuses, 71% also had exposure to nicotine, and 56% had exposure to other substances. Prior studies have shown lower brain size and volume in fetuses with prenatal nicotine exposure [
28,
29]. The multivariable models adjusted for nicotine exposure, but not for exposure to other substances, which also could have affected brain development. Fourth, fetal brain development could have been impacted by a range of additional factors such as maternal nutrition, stress, and environmental considerations. Fifth, because each fetus underwent a single MRI examination during the third trimester, we were unable to assess longitudinal brain development or identify a potential impact of in utero opioid exposure during earlier pregnancy time points. Sixth, although statistically significant differences between the two groups were observed for multiple biometric measurements, the differences were overall small. Moreover, the frontooccipital index was significantly different between the two groups, although the mean value was the same in the two groups based on the level of precision in this study. Seventh, we did not compare the two groups in terms of potential white matter change or other structural abnormalities, which could be assessed in future studies through diffusion-tensor imaging data. Finally, we did not assess associations of the prenatal findings with postnatal MRI or clinical outcomes. Such associations would help to further understand the clinical significance of the prenatal observations.
In conclusion, opioid-exposed fetuses, compared with unex-posed fetuses, had multiple smaller 2D biometric measurements of the brain on fetal MRI. In addition, fetuses with prenatal opioid exposure had increased frequencies of breech presentation and increased amniotic fluid volume. The findings indicate that prenatal opioid exposure contributes to a smaller in utero brain size and altered fetal physiology.