Anat Cell Biol 2020; 53(4): 451-459
Published online December 31, 2020
Copyright © Korean Association of ANATOMISTS.
1Department of Histology and Embryology, Tokyo Dental College, Tokyo, 2Department of Anatomy, Tokyo Dental College, Tokyo, 3Division of Internal Medicine, Jikou-kai Home Visit Clinic, Sapporo, Japan
Correspondence to:Masahito Yamamoto
Department of Anatomy, Tokyo Dental College, 2-9-18 Misaki-cho, Chiyoda-ku, Tokyo 101-0061, Japan
We have previously reported that the thoracolumbar posterior nerve root shows a tortuous epidural course, based on studies of human fetuses near term. For comparison with the cervical nerve, examinations were conducted using frontal, sagittal and horizontal sections of cervical vertebrae from 22 fetuses at 30-38 weeks of gestation. The cervical nerve root showed a short, straight and lateral course near the zygapophysial joint. Multiple rather than single bundles of the cervical posterior root seemed to account for the majority of sensory nerve fibers innervating the upper extremity. Fasciculation of rootlets was evident near the thoracolumbar spinal cord, whereas it was seen in the dural pocket at the nerve exit from the dural sac although both sites were subdural. As in the thoracolumbar region, the nerve sheath was continuous with the dura mater and independently surrounded each of the anterior and posterior roots. Radicular arteries were few in the cervical region. In 2 of the 22 fetuses (31 weeks and 33 weeks), there was a segmental, unilateral abnormality of nerve rootlet fasciculation where the dorsal root ganglion was located lateral or peripheral to the intervertebral region. Long nerve roots running inferiorly are a necessary adaptation to the delayed and marked growth of the thoracolumbar vertebral column. In children, the cervical nerve roots are likely to be affected by movement or dislocation of the vertebrae. The segmental abnormality of the cervical nerve root may be linked to rare variations in the brachial plexus.
Keywords: Spinal nerve, Fetuses, Anatomy
Based on gross observation of 20 adult specimens and histological studies, Tubbs et al.  described a region-specific difference in the architecture of the spinal nerve root sheath: the cervical anterior and posterior nerve roots were apparently separated, and each was surrounded by a dural sheath, in contrast to the common sheath shared by both nerve roots in the thoracolumbar region. They considered the arachnoid to be continuous with a reticular layer along the nerve roots peripheral to the pocket-like protrusion of the dura mater at the nerve exit. In view of the large volume of hard tissues that were likely present, the histological details described were those after dissection, and therefore as a result the topographical anatomy presented might have been altered.
At the thoracolumbar level, frontal sections demonstrate multiple spinal nerve roots with long longitudinal courses through the epidural space [2, 3]. Using frontal sections from human fetuses near term, we have previously shown that the posterior nerve root often has a tortuous epidural course resembling a string of beads at the thoracolumbar level , whereas a straight anterior root is usually accompanied by a radicular artery in this region (Fig. 1). The anterior and posterior nerve rootlets run inferolaterally, and as described by Tubbs et al. , each of them enters an independent pocket-like protrusion of the dura mater in the thoracolumbar region. In addition, surgeons and neurologists have noted that the dorsal root ganglion (DRG) of the second cervical nerve differs in location from that of the other cervical nerves .
Consequently, the aim of the present study was to characterize the cervical region-specific morphology of the spinal nerve roots, focusing on 1) the dural pocket at the nerve exit, 2) fasciculation of nerve rootlets by the sheath, and 3) the topographical relationship between the nerve root and the radicular artery. The sections we employed were prepared from the entire cervical vertebral column, with the intention of clarifying the actual topographical anatomy without dissection-related artifacts.
The study was performed in accordance with the provisions of the Declaration of Helsinki 1995 (as revised in Edinburgh 2000). We examined paraffin-embedded histological sections from 22 fetuses near term (gestational age approximately 30–38 weeks; crown-rump length [CRL] 250–315 mm). All fetuses were part of the large collection kept at the Department of Anatomy, Universidad Complutense, Madrid, and were the products of miscarriages and ectopic pregnancies at the Department of Obstetrics of the University. All sections were stained with hematoxylin and eosin (H&E). The sectional planes were frontal (3 fetuses), sagittal (12) and horizontal (7). The study was approved by Complutense University ethics committee (B08/374). Observations and photography were usually performed with a Nikon Eclipse 80.
Frontal sections clearly demonstrated a sac-like or bag-like appearance of the spinal dura mater that contained the cervical spinal cord (Fig. 2). In the present study we termed this the ‘dural sac’, although to various extents the dura mater continued to a sheath surrounding the nerve roots and DRG. The sac was thick and reached 0.1 mm at its thickest part (Fig. 3). The zygapophysial or facet joint was evident near the nerve roots. The joint cavity extended widely from anterior sections partly containing the vertebral body (Fig. 2A, B) to posterior sections including the posterior arch of the first vertebra (Fig. 2I). The epidural veins appeared to provide bilateral trunks, which were crossed or penetrated by the spinal nerve roots.
When the nerve root crossed and exited from the dural sac, its direction changed from inferior to horizontal: the angle at the exit was almost 120°, but sometimes less and nearly 90° (the anterior root in Fig. 3E). At the nerve exit from the dural sac, a pocket-like protrusion of the dura mater was evident at the lower cervical level (Figs. 2E, G, 3G, I). Higher (second and third) cervical nerve roots exited from the dural sac in more anterior sections (Fig. 2A, B), whereas in posterior sections the fifth and/or sixth nerve roots did so (Fig. 2F–I). Sometimes, the anterior and posterior roots crossed the sac together (Fig. 3D, G). The posterior root was usually composed of two to three bundles, in contrast to the anterior root, which was a single bundle (Fig. 3B, E, G). No nerve root exhibited a tortuous course after exiting from the dural sac. No radicular artery candidate with a longitudinal course was found in frontal sections. In addition, the DRG of the second cervical nerve was located specifically near the dural sac (Fig. 2C).
Since the nerve roots ran laterally rather than inferiorly in the cervical region, sagittal sections demonstrated nerve rootlet fasciculation more clearly than frontal sections (Figs. 4, 5). Fasciculation of nerve rootlets occurred in the dural pocket at the point where the nerve exited from the dural sac (Fig. 5A, E). The anterior nerve rootlets were almost always fasciculated into a single bundle at the nerve exit. In contrast, the posterior rootlets were likely to fasciculate into two or three bundles (Figs. 4E–G, 5E, F). However, multiple bundles of the posterior root were surrounded by a common sheath continuous with the dural sac. Irrespective of whether the posterior root was single or multiple, the anterior nerve root was located immediately anterior to the corresponding posterior root(s). During its long course, the anterior root was often separated from the other components without intermingling. This allowed us to trace it from the dural pocket (Fig. 5A, E) to its peripheral course along the DRG surface (Fig. 5D, G).
The dural sac pocket at the nerve exit (Fig. 5A, E) was larger than that seen in frontal sections (Fig. 3G, I), and thus it was longer anteroposteriorly than mediolaterally. No nerve root exhibited a tortuous course after exiting the dural sac. However, changes in nerve root direction were difficult to identify in sagittal sections. Epidural veins tended to run along the anterior, rather than posterior, wall of the vertebral canal (Figs. 4F, 5A). A few radicular artery candidates took a longitudinal course along the nerve root (Fig. 5C, G). The zygapophysial joint cavity appeared to extend widely from medial sections containing the spinal cord (Fig. 4E) to lateral sections including the DRG (Fig. 4H).
Horizontal sections clearly demonstrated a topographical relationship between the anterior and posterior nerve roots, the former being anterior to the latter and the two being separated by a narrow space (Fig. 6). Likewise, a continuation of the dura mater was easily traced from the pocket at the nerve exit to the sheath of the DRG. However, neither the nerve root nor its sheath was likely to contain the arachnoid. Moreover, in contrast to frontal and sagittal sections, we often found a radicular artery running along the anterior root (Fig. 6D). The zygapophysial joint and a vertebral arch were seen together in a horizontal section containing the DRG (Fig. 6A–C), the latter being located close to the superior articular process of the lower adjacent vertebra. At the joint, the lower vertebral element was located anteriorly and the upper vertebra posteriorly (Fig. 6C).
In two of the 22 fetuses examined, a segmental and unilateral abnormality of nerve rootlet fasciculation was observed: the sixth cervical nerve at 33 weeks (CRL 280 mm; Figs. 7, 8) and the seventh nerve at 31 weeks (CRL 260 mm). In these lower cervical nerves, rootlet fasciculation did not occur in the epidural space including the dural pocket but in the interradicular area containing the DRG. However, the fasciculation was loose and did not involve the dura mater. Moreover, the unilateral DRG of the segment was not present between the vertebrae but embedded in the scalenus muscles (Fig. 7I, J). Thus, the single DRG was located far anterolateral to the vertebral artery. In spite of the lack of fasciculation near the dural sac, the anterior and posterior roots were discriminated in lateral sections containing the vertebral artery, each of them being loosely aggregated (Fig. 8D). The vertebrae and radicular artery candidates showed no evident abnormality.
O’Rahilly et al.  were the first to examine the whole vertebral column of human fetuses, although the specimens were at relatively early stages (8 weeks). Castellana and Kósa  demonstrated cervical vertebral growth in late-stage human fetuses, but did not examine the whole column. The present frontal and sagittal sections appear to be the first to demonstrate the topographical anatomy of the whole cervical column, including the cervical nerves, at the late stage, and the features likely reflect the morphology at birth.
In comparison with our previous observations of the thoracolumbar nerve root in near-term fetuses Neural-dural transition at the thoracic and lumbar spinal nerve roots: a histological study of human late-stage fetuses., the cervical nerve root was characterized by a short, straight and lateral course immediately medial to the zygapophysial joint (Table 1). In the thoracolumbar region, a long inferior course was evident in the dural sac because rootlet fasciculation occurred near the spinal cord, and not in the pocket at the nerve exit. Although we had intended to seek structures in the cervical region analogous to those in the thoracolumbar posterior root, i.e., a long tortuous course resembling a string of beads (Fig. 1), no such course was found. Moreover, the distribution of the radicular artery appeared to be sparse in the cervical region. On the other hand, several common features shared by the cervical and thoracolumbar nerve roots were evident, i.e., 1) a pocket-like structure of the dural sac at the nerve exit and 2) a nerve sheath continuous with the dura mater and independently surrounding each of the anterior and posterior roots. Multiple rather than single cervical posterior root bundles appeared to account for the majority of sensory nerve fibers innervating the upper extremity.
The thoracolumbar nerve roots would likely be affected by mechanical stress caused by the marked difference in growth rates between the spinal cord and the vertebral column . Therefore, long nerve roots would be necessary for adaptation to the delayed and marked growth of the vertebral column, leading to development of the cauda equina . In contrast, because of their straight short course and location near the zygapophysial joint, the cervical nerve roots would likely be influenced by movement or dislocation of the vertebrae in children. Cattell and Filtzer  reported frequent anterior dislocation of a single cervical vertebra in childhood. Unlike the adult morphology , fetal vertebral bodies did not show a saddle-like arrangement providing anteroposterior stability. According to San Román et al. , this saddle-like arrangement appears in the late period of childhood.
In a study using horizontal sections from 11 fetuses at 10–24 weeks, Khorooshi et al.  reported that the cervical DRG was located at a level lateral to the narrowest part of the interradicular space, in contrast to the situation for thoracic DRG. We agree with their conclusion, with the exception of the DRG of the second cervical nerve. The thoracic DRG is located less than 1 mm from the dural sac, and therefore it is often adjacent to the dural pocket . In contrast, we found that the cervical DRG was never adjacent to the pocket in frontal sections and the dural sac was not included in sagittal sections showing the cervical DRG. Therefore, the epidural (or subdural) course of the posterior nerve root was long in the cervical (or thoracic) region. In this context, the highly lateral DRG observed here is especially noteworthy. Why was the degree of nerve root fasciculation related to the highly lateral DRG? Growth of the DRG might require some degree of nerve fiber ‘bundling’ along its course. At the lower lumbar and sacral level, the DRG is likely to exist within the dural sac after the nerve rootlets have shown a long inferior course . Therefore, although it is not critical whether the nerve course is subdural or epidural, nerve fiber bundling should occur proximal to the DRG.
Somite development is responsible for segmentation of the spinal nerves [11, 12]. Similarly, the dura mater seemed to be responsible for fasciculation of the spinal nerve roots, as here there was no continuity of the dura with the nerve sheath in unbundled roots with lateral shifting of the DRG. In fact, unilateral segmental anomalies of the vertebrae and ribs in relation to genetic background have been reported [13, 14]. However, the spinal dura shows no segmental border despite its origin from somites [15, 16]. Therefore, it is unlikely that any unilateral segmental abnormality of the dura would be attributable to redundancy of a specific gene. Conversely, we speculate that any small, regional or transient morphologic feature of the growing dura, such as folding, twisting or tearing, would be unlikely to affect the nerve root sheath, resulting in an unbundled nerve root. This is analogous to well-known failures of intestinal growth caused by neighboring structures, such as traction or entrapment by the amnion during physiological herniation , the peritoneum  and vitelline structures [19, 20]. Traction of the midgut by the leg is also likely .
We are grateful to Prof. Dr. Jose Francisco Rodríguez-Vázquez (Department of Anatomy, Universidad Complutense) for allowing us to use his materials.
Conceptualization: GM. Data acquisition: KK, MY, YH, NS, TM. Data analysis or interpretation: KK, NI. Drafting of the manuscript: HY, GM. Critical revision of the manuscript: SA. Approval of the final version of the manuscript: all authors.
No potential conflict of interest relevant to this article was reported.