OUP user menu

Retrograde Filling of a Thoracic Spinal Artery During Transforaminal Injection

Way Yin MD, Nikolai Bogduk MD, PhD, DSc
DOI: http://dx.doi.org/10.1111/j.1526-4637.2009.00629.x 689-692 First published online: 1 May 2009


Introduction. Transforaminal injection of steroids has been associated with infrequent but devastating neurological complications and death. Direct injection of particulate steroids into the vertebral artery or medullary arteries is the leading theory in the majority of these complications. Practice guidelines have been published to minimize the likelihood of directly encountering vulnerable arterial structures. However, retrograde flow into a vertebral or medullary artery has not been considered in the literature.

Case. This case demonstrates retrograde flow into a common arterial trunk with subsequent antegrade flow of intravenous contrast into a thoracic spinal artery during thoracic transforaminal injection.

Discussion. Antegrade flow of particulate steroids through direct cannulation of a vertebral or medullary artery has been advocated as one explanation for complications involving brain or spinal cord infarction. The possibility of retrograde flow into a common arterial trunk with subsequent antegrade flow into vulnerable arteries should also be considered as a possible mechanism by which embolic spinal cord or brain injury may occur.

Conclusion. Retrograde flow into medullary or vertebral arteries without direct cannulation can occur, and provides an alternative mechanism of potential injury to the spinal cord or brain during transforaminal injections.

  • Injection
  • Spine
  • Artery
  • Spinal Cord
  • Epidural
  • Steroid
  • Complication


Transforaminal injections of steroids have been implicated in causing serious neurological complications and death. Most incidents have occurred after cervical injections [1–7] but some have resulted from lumbar injections [7–9]. Although thoracic transforaminal injections are practiced [10], only one case report concerns a complication following T12-L1 injection [11].

Various conjectures have been raised to explain the mechanism of injury. Direct injection into a vertebral artery or medullary artery may occur if needles are placed ventrally in the intervertebral foramen. Circumstantial evidence currently favors arterial embolization of corticosteroid preparations as the mechanism of neurological injury. However, even when needles have been correctly placed dorsally in the foramen, inadvertent intra-arterial injection has been demonstrated in the course both of cervical and of lumbar transforaminal injections [6,12,13].

The assumption in the prevailing model of injury is that, when steroids are injected into an artery, the embolic material flows orthograde along the artery (i.e. downstream). Retrograde flow has not been considered in the literature, ostensibly because operators believe that arterial pressure would ensure downstream flow of any injectate. The following case demonstrates an example of retrograde flow into a common arterial trunk with secondary antegrade flow into a thoracic medullary artery. The phenomenon observed allows for a modified mechanism of injury involving reinforcing radicular arteries or vertebral arteries.


In a 35-year-old female with a left paracentral T6-7 disc protrusion and left-sided thoracic radicular pain, a left T6-7 transforaminal injection was undertaken using a posterior parasagittal approach following established guidelines [10]. Under multi-planar, real-time fluoroscopic guidance, the needle was placed dorsal, and very slightly caudal, to the dorsal root ganglion of the left T6 spinal nerve, just outside the intervertebral foramen (Figure 1a). Upon injection of a test-dose of contrast medium, appropriate peri-radicular spread was observed, but what appeared to be opacification of a thoracic spinal artery was evident on anterior-posterior fluoroscopy (Figure 1b). Digital subtraction angiography confirmed that this was an artery, with rapid pulsatile washout of injected contrast (Figure 1c).

Figure 1

Plain radiographs of a needle in position for a thoracic transforaminal injection, taken after the injection of a test-dose of contrast medium. (A) Lateral view. The needle is located in the T6-7 intervertebral foramen, with its tip lying over the dorsal and caudal aspect of the spinal nerve just distal to the dorsal root ganglion, circumferentially outlined by contrast medium (black arrow). (B) Anteroposterior view. Contrast medium outlines the spinal nerve within the intervertebral foramen, but also appears to fill a transversely running artery within the vertebral canal (black arrow). (C) Anteroposterior, digital subtraction angiogram of the image seen in B. The white arrow points to contrast medium outlining the spinal nerve, below the T6 pedicle. The black arrow points to what clearly is an intraspinal artery.

When digital subtraction angiography (DSA) was repeated, additional features emerged. On the lateral view, a small artery filled in a retrograde direction, starting from a caudal and dorsal position in the intervertebral foramen, and continuing ventrally across the lateral aspect of the vertebral body into an intercostal artery (Figure 2a). Subsequently, the contrast medium continued to opacify the intercostal artery, in a retrograde direction toward the thoracic aorta, but also filled a spinal artery in an orthograde direction toward the ventral and cephalad margin of the intervertebral foramen (Figure 2b). Subsequent lateral DSA imaging demonstrated filling of an artery in the ventral spinal canal, consistent with a medullary artery (Figure 2C). This latter vessel was distinctly separate from the original artery that was filled.

Figure 2

Sequential digital subtraction angiograms of a T6-7 transforaminal injection in lateral view. (A) An early view. From the caudal and dorsal aspect of the intervertebral foramen, opposite the tip of the needle, an artery (black arrow) fills in a retrograde direction into the intercostal artery lying on the lateral surface of the T6 vertebral body. (B) A subsequent view. The intercostal artery has back-filled as far anteriorly as the thoracic aorta, but contrast medium now flows in an antegrade direction into a spinal artery (black arrow) that enters the intervertebral foramen at its ventral and cephalad end. Contrast medium is absent in the original artery (white arrow) that first filled from the caudal aspect of the foramen. (C) A later view. Arterial filling with contrast medium can now be seen in the ventral spinal canal posterior to the T6 vertebral body (point of white triangle), corresponding with the expected course and location of a medullary artery. (D) A schematic illustrating the observed arterial flow pattern. Needle (black line) penetrates an artery (red) in the foramen dorsal to the spinal nerve (light gray oval). The foraminal artery (black arrow) is continuous with the segmental intercostal artery (black triangle). From the/intercostal artery, antegrade filling extends into a medullary artery, passing into the foramen ventral to the spinal nerve and also distally along its course in the ventral spinal canal, posterior to the T6 vertebral body (gray triangle).

The procedure was terminated, and the patient suffered no ill effects.


Dommisse has cautioned that the intervertebral foramina represent a “major arterial distribution point” and are of “critical value in the blood supply of the cord”[14]. Alleyne's microscopic dissections of the great medullary artery of Adamkiewicz illustrated arterial branches dorsal to the spinal nerves, continuous with the Artery of Adamkiewicz through a common trunk [15]. This case demonstrates that retrograde flow can occur within arteries following transforaminal injections. Furthermore, injectate can subsequently flow in an antegrade direction within a spinal or medullary artery that was not directly cannulated by the needle (Figure 2d). Although this phenomenon was observed at a thoracic level, it raises an intriguing possibility if applied to cervical levels.

It would seem feasible that a cervical transforaminal injection could back-fill a radicular (or other) artery and enter a vertebral artery. Subsequently, injectate could flow in an antegrade direction either into a nearby medullary branch of the vertebral artery or along the full course of the vertebral artery into the brainstem. If embolic material were so injected it could cause infarction of the spinal cord or of the brainstem. Embolic events in the territories supplied by the vertebral artery or a medullary artery, therefore, could occur without the artery itself actually having been punctured.

Physicians have already been urged to be alert to possible antegrade injection of a medullary artery or the vertebral artery [5,10,12,16]. The present case calls on them also to be alert to possible retrograde filling to a common arterial trunk.


View Abstract