1.3.2.1

Vascular damage

The neurologic impairment’s speed of onset argues in favour of the presence of a vascular injury. Medullary ischemia would explain the medullary hypersignals in MRI. However, the mechanism of this ischemia is not clear.

Below T8, the arteria radicularis magna (the major anterior radicular artery, also known as the artery of Adamkiewicz) constitutes the major blood supply to the anterior spinal artery . The artery of Adamkiewicz arises between T9 and L2 (from the left, in 85% of cases) in 85% of individuals and at the upper lumbar level in 10% of cases . However, its origin is highly variable; in a small proportion of subjects, the artery of Adamkiewicz may arise from near the upper vertebrae in the lumbar spine (or lower, near L3, L4 or L5) , the thoracic spine (T5 to T8, with a lumbar or sacral supply in 10 to 15% of cases) and, more rarely, from even as low as S1.

Low lumbar infiltration can damage an artery. The artery of Adamkiewicz is an enlarged, radicullomedullary artery which anastomoses with the anterior spinal artery; it extends as far as the lower part of the medulla spinalis and continues as a slender branchlet over the filum terminale. The anterior and posterior branches of the artery in the lumbar enlargement come together at the top of the conus medullaris (near L1 and L2) to form the latter’s anastomosing handle. Below the conus medullaris, the roots of the cauda equina (formed by the L2 to S5 roots) each receive a lower radiculomedullary artery; some terminate at the root, whereas others constitute a lower medullary supply which reaches the terminal conus medullaris (forming part of the latter’s anastomosing handle). This illustrates the vascular complexity of the cauda equina’s roots and the fragility of that of the conus medullaris (a single-end artery with practically no substitute). Hence, one can easily envisage that damage to the latter blood supply may cause a ventral medullary infarct and, as a result, complete paraplegia.

Compression by the injected volume is a poor hypothesis, given the lack of subsequent neurological recovery and the MRI images. Indeed, the liquid spreads out quickly or is quickly resorbed.

Might the needle have caused direct injury to an abnormally low, dominant radiculomeddulary artery (a known anatomical variant)? In support of this hypothesis, Sullivan et al. noted that the risk of intravascular uptake for foraminal infiltration is 10.8%. Contrast agent uptake by a medullary artery is frequently observed. However, preinjection aspiration failed to produce blood flashback in 74% of cases that proved to be intravascular. Hence, aspiration has very poor sensibility ; it should be performed but does not rule out intravascular injection.

One can suppose that the existence of associated vascular abnormalities increases the likelihood of vessel presence in the foramen. We cannot say whether or not our patient has a malformation which might make this type of complication more likely, such as reported by Quintero et al. (angiomatous malformation of T9). In the latter study, no lesions were seen in MRI but an arteriography was not performed.

So, what type of mechanism might be involved? Glaser and Falco suggested (by analogy with other arterial beds) that injury could result from an arterial spasm and/or the development of an intimal flap caused by a direct needle lesion. Both of these mechanisms could lead to stasis flow, clot formation, and hypoperfusion, as seen in coronary lesions . The diagnosis of this type of damage is problematic; arteriography could perhaps have provided more information in the present case.

Our case report describes a rare case of paraplegia following low spinal infiltration. The MRI images in two different areas of the spine suggested that embolization was more likely than a spasm or a direct arterial damage.

1.3.2.2

Causal links with corticosteroids

The Afssaps has emphasized that:

• •
  

  only cortivazol and prednisolone have received regulatory approval for epidural infiltration. Prednisolone is also authorized by intradural injection in the treatment of epidural-refractory sciatica and back pain;

  
  
• •
  

  other glucocorticoids (such as triamcinolone and methylprednisolone acetate) are only indicated for intra-articular, peri-articular and loco dolenti injections.

The latter two drugs were injected in five of the nine cases of medullary ischemia described in the literature.

Corticosteroid formulations contain microcrystals which may potentially occlude arterioles measuring 5–10 μm in diameter (such as the medullary vessels). To test this hypothesis, five commonly used preparations (methylprednisolone, triamcinolone, betamethasone sodium phosphate/betamethasone acetate and dexamethasone) were subjected to microscopic analysis. An unopened vial was shaken vigorously. The measurements were recorded and exported to a spreadsheet program. Measurements were sorted according to the particle size. We calculated the proportion of each size class as a percentage of the total number of particles. Aggregates (generally measuring between 1 and 10 μm) were found in each preparation. Some particles exceeded 50 μm in diameter (8.57%, 3.7%, 1.14%, 3.7% and 0% for the five drugs, respectively). The methylprednisolone and the triamcinolone even contained 100 μm particles. The diameter of the medullary arterioles is estimated to be between 10 and 15 microns, with the small medullary arteries measuring around 50 microns. Hence, embolization of a medullary artery by this type of particle could explain the damage described in the present case report. Although solutions of these steroid compounds are valuable because of their long-acting properties (from 36 to 72 hours), the risk may be greater than for suspensions .

Most of the reported cases involved steroid solutions but the present case and those reported by Quintero et al. and Lenoir et al. involved suspensions (a combination of three different compounds in Meyer et al. ).

Benzyl alcohol (the preservative agent in tiamcinolone) may have neurotoxic activity . The damage described in the various case reports does not argue in favour of such a phenomenon but these data suggest that caution should be taken when using these compounds.

1.3.2.3

Other factors

Although a history of spinal surgery is frequently described, it is not clear whether this factor has a role in the occurrence of paraplegia after injection .

The number of injections prior to an incident is often poorly reported in the literature. Some of the patients displayed the complication after one injection, whereas two or four injections were performed in other cases.

1.3.2.4

Clinical practice recommendations

Epidural infiltration is recommended when the patient presents subacute sciatica (regardless of the presence or absence of low back pain). The procedure is more rarely recommended for lumbago. Oral medication and physiotherapy are the recommended front-line treatments. Efficacy is better for recent-onset pain (1 to 60 days) than chronic pain (12 weeks to 1 year) .

There is no consensus on clinical practice recommendations for preventing these complications. Guidelines were initially proposed by Tiso et al. for cervical foraminal injections and were modified by Quintero et al. for lumbar foraminal infiltrations:

• •
  

  prefer a blunt-tipped needle, to reduce the risk of direct vessel lesions;

  
  
• •
  

  prefer a large-diameter needle (less than 22 G), to reduce the risk of cannulating a vessel;

  
  
• •
  

  perform a preinjection aspiration (it must be negative for cerebrospinal fluid and blood);

  
  
• •
  

  use the steroid with the fewest large particles (cortivazol and prednisolone have regulatory approval in France);

  
  
• •
  

  the use of contrast agent adds a margin safety by ensuring neural spreading and the lack of vascular uptake (check the position after handling the needle);

  
  
• •
  

  use microbore extension tubing to minimize needle handling after a confirmed placement. Medications are injected through the catheter, instead of through the needle. Multiple-level injections require the use of multiple microbore extensions;

  
  
• •
  

  inject a local anaesthetic “test” (e.g. adrenaline-free lidocaine) and allow time to pass (time not defined) to detect the occurrence of neurological signs before injection of the corticosteroid (hence the importance of not using sedatives);

  
  
• •
  

  inject the corticosteroid slowly. Check regularly for the absence of blood in the aspirate.

These clinical practice recommendations are empirical and result from previously described observations. The various ways of performing the infiltrations have never been compared in a clinical study. Meyer et al. pointed out that some spinal injuries have been described after injection of anaesthetics, whereas this type of drug is given before corticosteroids. Adrenaline-containing formulations must not be used.

The sacrococcygeal route may be of value: there is less risk of damage to the blood vessels in general and major anterior radicular artery in particular. This procedure is not mentioned in the indications.