Comparison between MRI and the Combination of 2D and 3D US in the Prenatal Diagnosis of Closed Spina Bifida

Spina bifida is a common congenital nervous system disorder caused by incomplete closure of the neural tube at 22–32 weeks of gestation. It accounts for approximately half of all cases of neural tube defects [1]. Closed spina bifida (CSB) refers to a spinal anomaly that is covered with skin, is unrelated to hydrocephalus or hindbrain herniation, and may or may not include back mass [2]. CSB affects the lumbosacral region, accounting for 10–15% of spinal abnormalities; its incidence stands at 0.1–0.14/1,000 [3, 4]. Notably, CSB is used as a general term to cover the tethered spinal cord, lipomyelomeningocele, meningocele, and diastematomyelia (split spinal cord) [4]. Failure of the neural and mesodermal tissues to differentiate and separate from each other appears to be the primary neural tube defect in CSB. Consequently, the conus medulla or cauda equina adheres to the posterior wall of the spinal canal, resulting in tethered spinal cord syndrome, a disorder of secondary neurulation [5]. Generally, the outcome for infants with CSB is desirable; however, neurologic symptoms of variable entity are common [6]. They include neurogenic lower urinary tract dysfunction, including urinary incontinence and multiple urinary tract infections; musculoskeletal disorders including movement disorders; bowel dysfunction such as constipation and fecal incontinence; sexual dysfunction including erectile dysfunction and dyspareunia; pain; and social problems.

CSB is prenatally difficult to diagnose, with common cases of misdiagnoses. Vertebral arch affected with skin overlapping the defect is a visible abnormality under ultrasound. In CSB, meninges protrude through the meningocele, causing no cerebrospinal fluid leaks [7]. Also, the cranial anatomy is normal. Since the neural tube defect is covered by skin, there is no cerebrospinal fluid loss, nor any herniation [8]. During the first-trimester routine examination for spina bifida, Zhu et al. [9] evaluated cranial ultrasound markers. Consequently, they found good ultrasound markers for the diagnosis of open spina bifida. However, these ultrasound markers should not be used in CSB.

Of note, magnetic resonance imaging (MRI) has important clinical value in prenatal diagnosis of spina bifida. Nevertheless, it is expensive, time-consuming, and affected by maternal breathing or fetal movement. Recent studies used 2D US combined with the 3D US to detect spina bifida occulta and reported accurate diagnostic capacity by recognizing the position of the conus medullaris [4, 10, 11]. Nonetheless, systematic markers for CSB diagnosis remain unavailable.

Therefore, this retrospective study aims to analyze the application value of prenatal 2D US complemented with 3D US and MRI in the diagnosis of fetal CSB. This is geared towards understanding the clinical features of CSB.

This retrospective study recruited 20 fetuses diagnosed with CSB which was verified by postnatal MRI, post-mortem pathological examination, or postpartum surgery between February 2016 and October 2021 at the First Affiliated Hospital of Nanchang University. We performed prenatal ultrasound and MRI in all cases. This study was approved by the Medical Research Ethics Committee of the First Affiliated Hospital of Nanchang University. All the women provided written informed consent for the participation of fetuses.

The inclusion criteria included (1) fetuses diagnosed with CSB which was verified by postnatal MRI, post-mortem pathological examination, or postpartum surgery; (2) all fetuses underwent prenatal 2D and 3D ultrasound and MRI. The exclusion criteria included (1) clinical information about pregnant women was missing; (2) image quality of the prenatal ultrasound and MRI images did not meet the standards for analysis.

Pregnant women were asked to lie in the supine position. Prenatal 2D and 3D ultrasound examinations were performed using a transabdominal 2D convex array probe with a frequency of 2–7 MHz, and a 3D volume probe with a frequency of 5–8 MHz (GE Voluson E8/E10; General Electric, Schenectady, NY, USA). The IU22 diagnostic Doppler ultrasound system with a 5–12 MHz transducer (Philips, Amsterdam, Netherlands) was used for induced labor fetuses. The fetal system and appendages were comprehensively scanned as per ultrasound specifications for fetal system scans, focusing on the fetal spine, conus medullaris position, and integrity of the skin as well as soft tissues on the fetal spine surface. The major points of spine examination include multisection scanning of the fetal spine, i.e., sagittal section, cross-section, and coronal section of the spine; observation of the position of the medullary cone; and the combined use of results from all sections to identify and judge the shape and structure of the spine and the relationship between surrounding tissues. The 2D and 3D US radiologists who carried out the scans and outcome interpretation had more than 5 years of clinical experience.

3D ultrasound volumes were obtained trans-abdominally from sagittal images of the fetal spine with the fetal back facing upwards; the lumbosacral area was selected as the area of interest. The acquisition angle was 60°, and the acquisition time was 4–10 s with a “maximum” energy modem. The obtained images were stored in the in-built hardware of the machine for subsequent post-processing image analysis under different methods.

Coronal plane acquisition: the omni-view mode of 3D ultrasound volume contrast imaging (VCI) was used to draw curves along the vertebral body and vertebral arch on the sagittal section of the spine. The coronal image of the vertebral body and vertebral arch was obtained by adjusting the curvature and angle of the line. The first image obtained was the coronal plane (C plane). At the same time, the C plane and the sagittal plane (A plane) were displayed on the same screen, and plane C was partially amplified. The condition of the spine was observed using multiple planes.

Position of the conus medullaris: 3D volume data analysis was conducted to locate the conus medullaris. The 3D ultrasound VCI mode was selected, with a layer thickness of 1–3 mm, and we adjusted three mutually perpendicular spine sections. Plane A was adjusted to the mid-sagittal section of the spine, revealing a slight lumbar spinal cord. The end of the enlarged spinal cord was the conus medullaris, which was the center point. At this time, plane B was the transverse section of the corresponding vertebral body, and plane C was the coronal section of the spine perpendicular to planes A and B. Eventually, the thickness of the VCI mode was adjusted to 4–8 mm; therefore, plane C revealed the ribs and vertebral body. The vertebral body corresponding to the lowermost rib was the reference vertebral body, i.e., the 12th thoracic vertebra T12. We recorded the vertebral body corresponding to the end of the conus medullaris.

Ultrasound imaging was used to observe the major images of various types of CSB, including spine curvature in the longitudinal section; the asymmetry between two rows of vertebral arches in the coronal section (segmental vertebral dysplasia); lack of clarity in the conus in the spinal canal or the sacral location of the conus (filamentary tension); the presence of an intraspinal or sacrococcygeal subcutaneous hyperechoic mass (intradural or intraspinal lipoma, or terminal filament lipoma); the connection between the back or sacrococcygeal subcutaneous nodule soft tissue to the spinal canal tissue (fur finus); a short or absent caudal vertebral body (caudal degeneration syndrome).

MRI was performed with a 3.0T superconducting magnet (Trio TIM Magnetom, Siemens Medical Systems, Erlangen, Germany). Gradient field strength was 45 mT/ms, and we used an 18-channel phased array body surface coil. Pregnant women were instructed to lie in the supine position. The three orthogonal planes of the middle and lower abdomen of pregnant women were scanned, before the fetal target spine. Eventually, the suspected fetal spine and spinal cord abnormalities by ultrasound were set as the center; the Haste, Trufi, SWI, and T1 2D fast small-angle excitation (Flash) sequence scans were performed in the sagittal, coronal, and axial positions.

The description and diagnosis of fetal MRI and ultrasound were performed by two physicians with more than 5 years of MRI and ultrasound diagnosis experience in obstetrics and neonatology. In case of disagreement, a decision was made by collective consultation of the radiology department. The results of MRI and ultrasound were blind to each other.

The presence of any two of the four criteria mentioned below was considered a positive diagnosis of CSB: (1) the position of the conus below the L3 level; (2) the interpediculate distance was greater 95% than the vertebral body width; (3) the presence of scoliosis; (4) the presence of vertebral arch fissure.

Statistical data were analyzed using the Statistical Package for Social Sciences (SPSS) version 17.0 (SPSS, NY, USA). Continuous data were expressed as median (range), and categorical data were presented as n (%). Diagnosis, missed diagnosis, and misdiagnosis rates of 2D combined with 3D ultrasound and MRI examination were calculated to evaluate the diagnostic efficiency. The κ statistic test was calculated to correlate the characteristics detected by ultrasound and MRI. A κ value of 0–0.4 indicated marginal agreement, 0.41–0.75 good agreement, and 0.76–1.00 excellent agreement.

We screened 23 fetuses for eligibility, of whom 3 fetuses were excluded due to the prenatal ultrasound and MRI images that did not meet the standards for analysis. Overall, this study obtained 20 fetuses with postpartum-diagnosed CSB. They include 6 cases confirmed by post-mortem pathological examination, 1 case confirmed by postpartum surgery, 13 cases detected by postnatal MRI. Of note, all the pregnancies were singleton. The average maternal age was 25 (20–42) years, BMI ranged from 22 to 26 kg/m², and gestational age ranged from 22 to 32 weeks (Table 1). Most of the pregnant women (12 [60%]) had a history of adverse pregnancy, i.e., 5 cases of embryo suspension, 2 cases of biochemical pregnancy, 1 case of fetal malformation, 3 cases of fever in 1st trimester, and 1 case of massive bleeding during pregnancy.

Among all the cases, 3 (15%) were meningocele, 6 (30%) were lipomas, 1 (5%) was lipomyeloschisis, 1 (5%) was terminal myelocystocele, 1 (5%) was spinal teratoma, 2 (10%) were diastematomyelia, 3 (15%) were dermal sinus, 1 (5%) was lipoma of the filum, 1 (5%) was lipoma of the conus, and 1 (5%) was caudal degenerative syndrome. Regarding deformity type, 11 (55%) had single malformations, whereas 9 (45%) had multiple malformations, out of which 4 (20%) had combined cardiac malformations, 1 (5%) had craniocerebral abnormalities, 2 (10%) had cleft lip and palate, 2 (10%) had limb malformations, and 1 (5%) had other malformations. Lesions were located in the lumbosacral vertebrae in 18 (90%) of fetuses diagnosed with CSB and in the thoracic vertebrae in 2 (10%) of the fetuses.

Regarding the pregnancy outcome, 16 (80%) fetuses were subjected to induced abortion, and 4 (20%) were delivered to term. The birth was followed up for 1 (0.5–4) years. One of the 4 born infants developed lower limb muscle weakness, neurocystitis, and other related symptoms, and was subjected to surgery in the first year of birth. The other 3 cases had no remarkable symptoms.

A total of 14 (70%) fetuses were diagnosed by 2D and 3D US, with 13 cases (45%) where the conus medullaris was lower than L3; 4 (10%) cases were diagnosed with vertebral arch fissure, 11 (20%) had interpediculate distance ≥95%, and 9 (25%) had scoliosis (Table 2). The misdiagnosis and missed diagnosis rates were 15.0% (3/20), and 15.0% (3/20), respectively.

We confirmed 10 (50.0%) cases with back protrusion, where the vertebral body was split left and right, showing a “U” or “V” shape, among which 1 case showed a large mass of about 54 mm × 67 mm, resulting in signs of cerebellar tonsil herniation in the skull, with “lemon sign” and “banana sign.” Additionally, we confirmed 4 (20.0%) cases without back protrusion, among which 2 (10.0%) case showed a suspicious hyperechoic space in the spinal canal and 2 (10.0%) case with subcutaneous sinus. Further, 3 (15.0%) cases were misdiagnosed, i.e., 1 case was misdiagnosed as hemivertebra; 1 as vertebral malformation; and 1 as a vertebral arch developmental disorder.

A total of 15 fetuses were diagnosed by MRI, with a diagnosis rate of 75.0%; 14 (70%) exhibited conus medullaris lower than L3; 7 (35%) cases had vertebral arch fissure; 7 (35%) cases had interpediculate distance ≥95%; 7 (35%) cases had scoliosis (Table 2). We observed 11 cases of CSB with protrusion showing sac-like masses and the contents of the spinal cord as well as meninges. Moreover, we observed 4 cases of spina bifida occulta, among which 2 revealed abnormal ossification signals in the spinal canal, dividing the spinal cord into two; 1 case showed a subcutaneous sinus connected to the spinal canal, and the sinus was twisted; 1 case showed a short T1 and short T2 signal at the end of the vertebral spinal cord; the fat suppression image revealed low signal. Additionally, 20.0% of fetuses (4/20) were misdiagnosed as congenital scoliosis, i.e., 1 case was misdiagnosed as a hemivertebral deformity, 2 as vertebral malformation, and 1 as segmentation failure. In 1 (5.0%) case, both prenatal ultrasound and MRI missed the diagnosis of spina bifida but only diagnosed the intraspinal occupying space with low spinal dysraphism at level S4. This case was postoperatively confirmed as CSB with intraspinal teratoma (Fig. 1).

The diagnosis accuracy of 2D and 3D US was comparable with MRI (70% vs. 75%, κ = 0.62); 2D and 3D US detected more cases with interpediculate distance ≥95% (55% vs. 35%, κ = 0.22) than MRI. On the other hand, MRI had a better capacity for identifying vertebral arch fissures (20% vs. 35%, κ = 0.39). MRI and 2D + 3D US had good agreement in conus medullaris (65% vs. 70%, κ = 0.42) and scoliosis (45% vs. 35%, κ = 0.59). Both 2D with 3D US and MRI detected 1 (5.0%) case with “lemon sign” and “banana sign.” The missed diagnosis rates of 2D combined with 3D US and MRI were 15% (3/20) and 5% (1/20), respectively, whereas the misdiagnosis rates of 2D with 3D US and MRI were 15.0% (3/20) and 20.0% (4/20), respectively.

Unlike open spina bifida, CSB is the insufficiency of the spine with continuous and complete skin at the lesion. It is more likely to occur in the sacrococcygeal region and causes tethered cord syndrome, resulting in varying degrees of neurological dysfunction after birth. During prenatal diagnosis, CSB is difficult to detect with its misdiagnoses and missed diagnoses commonly encountered in clinical settings. This poses a significant problem in China [12, 13]. This study analyzed 20 cases of CSB and detected major signs on imaging, including conus medullaris lower than L3, vertebral arch fissure, interpediculate distance, and scoliosis, along with back protrusion connected with the spinal canal and a split in the vertebral body. The diagnosis rate of fetal CSB by combined 2D and 3D US was 70%, whereas that by MRI was 75%. This indicates comparable efficacies of combined US and MRI in prenatal detection of CSB.

The fetal spine can be examined by US using multiple slices, including the sagittal, transverse, and coronal sections of the spine, to identify its structure and relationship with surrounding tissues, before providing the primary and the most precise mode of prenatal diagnosis [7]. Because of the complicated classification, CSB is classified into spina bifida occulta or other types of CSB with a protrusion, including lipomyeloschisis, meningocele, and thickened filum terminale [12]. The presence of neural tissue within the mass or sac is often observed, although US cannot reliably exclude the neural tissue [5]. The mass is connected to the spinal canal, and the ossification center of two vertebral arches located at the back can be observed, hence revealing typical “V”- and “U”-shaped changes. Cystic mass features associated with spinal dysraphism are often anechoic with a thin wall of open defects. On the other hand, cystic masses associated with closed defects have a thick wall and/or a complex appearance with echogenic components [8].

One previous study revealed 4 closed defects in 66 cases of fetal spina bifida diagnosed at a median gestational age of 21 (16–34) weeks. Among these, 2 (50%) had a posterior cystic mass with thick walls and a complex appearance, whereas the spinal lesion could not be differentiated from an open defect in 2 (50%) cases [8]. In this work, 10 cases (10/20) of mass-type CSB were detected at 22–32 weeks of gestational age by changes in 2D and 3D ultrasonographic images, demonstrating a “U” or “V” shape. Due to a large mass, 1 case had signs of cerebellar tonsil hernia. A cerebellar tonsil hernia or an Arnold-Chiari type II malformation may occur if the anechoic cyst is too large [14]. Our findings revealed a higher diagnostic rate by late gestation than that by previous findings. We considered the presence of scoliosis a positive diagnostic indicator of CSB. Non-mass CSB had smaller splits at the vertebral arch. Predicting the spinal level of the lesion is critical in determining prognosis [5]. A systematic review and meta-analysis confirmed that ultrasound and MRI can identify the upper level of the lesion in fetuses with open spina bifida (40.9% vs. 42.5%), leading to moderate diagnostic accuracy [15]. In another case series of 84 cases of open spina bifida, 3D ultrasound diagnosed a 70% lesion level of spinal segment in the last trimester [16]. Only 4 of 10 cases of CSB without mass were diagnosed using ultrasound. Among these, 2 case of abnormally strong echo was in the spinal canal, and 2 case of subcutaneous sinus. We encountered difficulties in prenatal ultrasound diagnosis of CSB without mass. Besides, 2D and 3D ultrasound scans underestimated the true lesion level, specifically by community obstetric sonography [17‒19]. Furthermore, combined 2D and 3D ultrasound identified 4 (20%) cases with vertebral arch fissures and 11 (55%) cases with interpediculate distance ≥95%. An Interpediculate distance of ≥95% was considered a positive diagnostic indicator of CSB.

When diagnosing CSB, the position of the conus medullaris is an important point of focus in spinal cord assessment. Sattar et al. [20] evaluated fetuses with CSB and discovered a long spinal cord as the commonest structural abnormality. As a consequence, they evaluated the vertebral level of the conus medullaris in all fetuses, specifically in patients with a history of spina bifida. Elsewhere, 3D ultrasound accurately established the position of the conus medullaris to detect spina bifida occulta. Notably, 3D ultrasound has been adopted with volume contrast imaging to further establish its diagnostic capacity, resulting in confirmed accuracy in fetal spinal cord assessment [10]. Milani et al. [4] combined both 2D and 3D ultrasound in prenatal diagnosis of 6 fetuses and noted that the conus medullaris with a position lower than expected for gestational age and posterior disposal in the spinal canal increases the suspicion of possible CSB. Similarly, we considered the position of the conus below the L3 level as a positive diagnosis of CSB.

Generally, typical changes in cranial ultrasound images are lacking in fetuses with CSB. Retrospectively, Bahlmann et al. [11] analyzed fetuses with spina bifida using 2D ultrasound and discovered the lemon and banana signs in standard measuring planes as indirect ultrasound markers of spina bifida. Nevertheless, only 1.1% of the cases had CSB, and the lemon, as well as banana signs, were commonly detected before 24 weeks of gestation. In later stages of pregnancy, the cerebellum cannot be detected on ultrasound in about 80% of affected fetuses [5]. We encountered only 1 case with lemon and banana signs in late gestation. Therefore, the lemon and banana signs are unreliable indicators of CSB.

With the advancement in clinical application value of fetal MRI, the benefits of MRI in detecting fetal spine and spinal cord dysplasia have become increasingly paramount. The advantages of MRI include high spatial resolution and soft tissue resolution; also, the image quality is unaffected by weight, amniotic fluid volume, gas in the imaging region, and bone influence. MRI imaging of CSB fetuses revealed continuous and uniform signals of soft tissues and skin at the back of the spine, indicating enlarged spinal canal and sac-like masses, with signs of bulging of the spinal cord, meninges, and their contents. Thus, MRI reveals the structure and pathological changes of the fetal spine and spinal cord, which can to a certain extent compensate for the inadequacies of ultrasound [21, 22].

We diagnosed 2 case with an abnormally strong echo in the spinal canal. This was confirmed by MRI as a longitudinal spinal cord caused by a bone crest. Moreover, results showed that the diagnostic accuracy of 2D and 3D US was comparable with MRI (70% vs. 75%, κ = 0.62). This may be attributed to the poor diagnostic capacity of vertebral arch fissure by US (20%) in marginal agreement (κ = 0.39), and poor diagnostic ability of interpediculate distance ≥95% by MRI (35%) in marginal agreement (κ = 0.22). Nevertheless, MRI and ultrasound corroborated the conus medullaris (κ = 0.42) and scoliosis (κ = 0.59). These findings are in line with that of previous studies. Brandon et al. [23]. Compared the accuracy of fetal MRI and prenatal ultrasound in predicting lesion levels in patients with myelomeningocele and discovered that prenatal MRI and US agreed within one spinal level in 74% of cases with a κ of 0.43. Araujo Júnior et al. [24] compared the use of 2D ultrasound and MRI in the diagnosis of 15 fetuses with spina bifida (encephalocele, rachischisis, myelomeningocele). Consequently, no significant differences were observed in atrium sizes of the lateral ventricle and first vertebra between both imaging techniques. MRI examinations for CSB without masses have a number of benefits. Due to changes in fetal movement and spine physiological curvature, ultrasound enables dynamic observation of the fetal spine. The 4 misdiagnoses by MRI suggested its shortcomings in the display of vertebral deformities.

This study has worth-mentioning limitations. First is its retrospective nature. Secondly, its small sample size may limit the generalizability of the results. Additional prospective multicenter studies with a larger sample size are necessary to validate our findings. Thirdly, fetal MRI is a further assessment when the prenatal ultrasound is suspected to be positive, which has a certain effect on the diagnostic rate.

In conclusion, 2D combined with 3D ultrasound remains the first choice for CSB diagnosis. Nonetheless, it is vulnerable to factors including fetal position. Due to its high soft tissue resolution, MRI has benefits in the prenatal detection of spinal cord lesions. Combining conventional 2D and 3D ultrasound yields a similar diagnostic rate of CSB to MRI. Therefore, vertebral arch space and scoliosis detected by ultrasound may help improve the detection and diagnosis of CSB.

This study protocol was reviewed and approved by the Local Medical Ethics Committee of the First Affiliated Hospital of Nanchang University (No. [2022]CDYFYYLK[09-025]). All the women provided written informed consent for the participation of fetuses.

The authors have no conflicts of interest to declare.

This work was supported by the Health Commission Science and Technology Plan of Jiangxi Province (No. SKJP202210452).

Weiping Zhang and Li Chen designed project and wrote the manuscript. Cases were collected and followed up by Jingling Wang. Hui Wu analyzed the data. All authors reviewed the results and approved the final version of the manuscript.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.