Objective. Reports of catastrophic neurologic injuries following lumbar transforaminal epidural steroid injections are rare but serious potential complications. The traditional method of performing lumbar transforaminal epidural steroid injections is in the “safe triangle” to avoid contact to the spinal nerve. Some authors advocate an alternative approach by placing the needle inferiorly in a region referred to as “Kambin's triangle” to avoid incurring arteries. This study aimed to determine the location of arteries within the L1–L4 intervertebral foramen in vivo, specifically if they lie within or in close proximity to the “safe triangle” or Kambin's triangle using CT angiograms of the abdomen and pelvis.
Study design. The authors retrospectively evaluated the location in vivo of arterial vessels in the intervertebral foramen from L1 to L4 in patients who underwent abdominopelvic CT angiograms for aortic vascular disease. The data were reanalyzed to confirm inter-rater reliability.
Results. Arteries were found in both the safe triangle and Kambin's triangle at a statistically significant rate ( P < 0.05).
Conclusions. In this group of patients, an artery was found in either the safe triangle or in Kambin's triangle frequently, suggesting the location of these arteries can be quite variable. Physicians performing these procedures should use universal precautions to avoid inadvertent injection into the lumbar spinal arteries and minimize potential complications regardless of the approach.
Lumbar transforaminal epidural steroid injections (L-TFESIs) are commonly used as a treatment option in lumbar radiculopathy and lumbar spinal stenosis [ 1–3 ]. The traditional approach for a L-TFESI is a subpedicular placement of the needle in a region commonly referred to as the “safe triangle.” When this technique was originally developed, it was to avoid needle placement into the spinal nerve only and did not take into account the arteries found in the intervertebral foramen.
A rich anastomotic network of arteries provides circulation to the lumbar spine and the cauda equina. Dommisse et al. have shown that the spinal cord is essentially reliant on the anterior longitudinal artery of the spinal cord and the posterolateral longitudinal arteries of the spinal cord [ 4 ]. These arteries are fed by a number of medullary arteries that appear and arise at various levels, the largest of which is the artery of Adamkewicz. The lumbar spinal arteries arise from the aorta at levels L1–L4 and the median sacral artery at L5 [ 5 ]. The lumbar spinal arteries divide at the intervertebral foramen and send off branches to supply nutrients to the vertebrae and anastomose with the radicular arteries that receive additional blood supply from arteries from the conus medullaris. This anastomotic network no doubt provides a way for arterial flow to be maintained under conditions of stress [ 6 ]. This area has been described by Dommisse and Parke et al. as a zone of relative hypovascularity as the radicular arteries in this region are potentially receiving blood from the longitudinal arteries and the lumbar spinal arteries [ 7 , 8 ].
The typical mechanism of spinal cord infarction secondary to L-TFESI is thought to be direct injection of particulate steroid into a major medullary artery. However, there has been a case report describing retrograde flow in a intercostal artery following arterial injection of a thoracic TFESI into a common arterial trunk, which could be a possible mechanism of spinal cord injury [ 9 , 10 ]. It is clear, that the lumbar spinal arteries or any artery that it branches off to in the intervertebral foramen has the potential to communicate with one of the medullary arteries and in particular the artery of Adamkewicz with either antegrade or retrograde flow.
Bogduk et al. have cited a number of potential complications from L-TFESI ranging from minor problems such as headaches, worsening back pain, flushing, increased leg pain, vasovagal reaction, to major complications such as spinal cord infarction [ 10–17 ]. Other complications of L-TFESI include perineural hematoma [ 18 ], epidural hematoma [ 16 , 17 , 19 ], and infection [ 13–15 , 20 ].
As previously stated, the presumed mechanism of spinal cord infarction secondary to L-TFESI is the occlusion of a medullary artery with particulate steroid that communicates with the longitudinal arteries of the spinal cord. Vascular penetration during L-TFESIs using the traditional safe triangle approach is estimated to be between 6.1% and 15.5% [ 12 , 21–24 ]. It is when there is arterial penetration that there is potential communication with a major medullary artery including the artery of Adamkiewicz (AKA) that a potential embolic spinal cord infarction may occur.
Recently, the safety of the subpedicular approach has been called into question as this route is via the safe triangle. It is thought that lumbar spinal arteries or a communicating branch may be inadvertently penetrated and injected. A study of cadavers also described arterial vessels located in the superior aspect of the intervertebral foramen within the region of the “safe triangle” [ 25 ]. A study using angiograms in the thoracic and lumbar spine demonstrated the majority of arteries of the AKA located in the superior aspect of the intervertebral foramen, although also present in the inferior aspect 2% of the time [ 26 ]. This study was limited in that lateral images were not readily available and the interpretations were made only using posterior–anterior images.
As a result of the questioned safety of the subpedicular approach, many clinicians have advocated the use of a infraneural technique with needle placement in Kambin's triangle when performing L-TFESIs, arguing the likelihood of arterial penetration and injection will be less likely with this approach” [ 11 , 27–30 ].
While spinal cord infarction is a potentially devastating complication of L-TFESIs, the incidence is rare, even with the widespread use of particulate steroids [ 12 , 15 , 23 , 31 , 32 ]. The inferior approach also carries a possible risk of intradiscal injection [ 30 ]. Case reports of vascular flow during discography with the inferior approach are reported [ 33 , 34 ]. In addition, personal experience and informal polling of physicians within our group and at conferences noted occasional vascular flow with the inferior approach. A letter to the editor from Simon et al. describes two cases of this, but the flow appeared to be venous [ 35 ].
The purpose of this study is to determine the location of arterial vessels with respect to the lumbar intervertebral foramen by a retrospective analysis of computed tomography (CT) angiograms of the abdomen and pelvis that were obtained for evaluation of aortic vascular disease. Of particular interest is whether these arteries lie in a region that could be cannulized in performing a L-TFESI. In other words, could these arteries potentially be punctured while performing a L-TFESI? The location of these arteries with respect to the safe triangle and Kambin's triangle were also studied.
Institutional Review Board approval was obtained from our institution prior to the study. Thirty-two patients were enrolled for the study for a total of 256 vertebrae studied. Enrolled patients underwent CT-angiograms of the abdomen and pelvis between 2011 and 2013 (Toshiba Aquilion 16 or 64 system, determined by CT scanner availability, with 100 mL of Omnipaque (350 mg/mL) nonionic contrast (GE Healthcare Medical Diagnostics, Princeton, NJ) given as a dynamic bolus at 4 mL/s with bolus tracking in the arterial phase of the CTA protocol and these were retrospectively reviewed. All CTA images were reformatted (16 slice, 1 mm intervals; 64 slice, 0.75 mm intervals) in the axial plane and three-dimensional volumetric reconstructions performed using the Siemens AG Leonardo Syngo MMWP VE 25A Workstation (Siemens, New York, NY). Patients with evidence of prior lumbar surgery were excluded. It was determined if the arteries resided in the regions about the safe triangle, the inferior triangle (Kambin's Triangle), both or neither at each level and on each side between lumbar vertebrae L1–L4. The authors assigned these independently and then reviewed each case together to assure inter-rater reliability. Examples of images where the arteries can be found are in Figure 1 .
After the data were collected, the overall incidence that an artery was found in either the safe triangle or in Kambin's triangle was calculated for each level L1 through L4 with the value for standard error. The P value in this study represents the probability that an artery is found through random chance in either the safe triangle or in Kambin's triangle and that there is no relationship between vertebra level and location of an artery. A chi-square test was used to compare the vertebrae levels through a conditional logistic regression, which was then used to calculate the P values.
|Safe Triangle: 129/256||Safe||% Safe||Std Err|
|Kambin/inferior triangle: 52/256||Inferior||%Inf||Std err|
|Safe Triangle: 129/256||Safe||% Safe||Std Err|
|Kambin/inferior triangle: 52/256||Inferior||%Inf||Std err|
An artery was found in the safe triangle in 129 out of 256 vertebrae studied at a rate of 50.39%. The incidence an artery was found in the safe triangle in our study peaked at L2 with a rate of 64.06% and was lowest at the L4 level with a rate of 26.56%. The P value suggesting the arteries are found in the safe triangle randomly was 0.00008038.
An artery was found in Kambin's triangle in 52 out of the 256 vertebrae studied with an overall incidence of 20.31%. The incidence an artery was found in Kambin's triangle peaked at the L3 level at 29.69% and was lowest at the L1 level at 7.81%. The P value suggesting the arteries are found in Kambin's triangle randomly was 0.008415.
Interestingly, an artery was found in both the superior and inferior access points in 21/256 vertebrae for an overall incidence of 8.2% ( Figure 1 ).
Human arterial anatomy is known to be quite variable with common examples including the Circle of Willis in the brain, the location of the artery of Adamkewicz, and mesenteric arterial vasculature [ 36–39 ]. Studies have shown variability in the vasculature in the cervical spine, as well [ 40 , 41 ]. Dommisse also showed medullary arterial variability from cadaveric dissections and stated that it was a “striking feature” of the dissections he performed [ 4 ]. He also stated that the network which he refers to as the “arterial circle” is “more complex than the Circle of Willis” [ 4 ].
A study from Barrey et al. showed significant arterial variability in the spine in preoperative lumbar spinal CT-angiograms [ 42 ]. It should not be surprising that the location of the radicular arteries can also be variable in the lumbar spine and can occur in the safe triangle, Kambin's triangle or in both.
In our study, an artery was found rather frequently in either the region of the safe triangle or in Kambin's triangle with an overall incidence of 50.39% and 20.31%, respectively. The data also support previous studies that arteries are more likely to reside in the region of the safe triangle. Furthermore, the P values suggesting the arteries occurred in either location randomly is rather low, suggesting that there is a strong relationship with location of these arteries occurring in the safe triangle and in Kambin's triangle in the lumbar vertebrae.
Recent evidence has made an argument for the use of nonparticulate steroids in L-TFESIs. The rationale is that the particle size in dexamethasone is smaller than would potentially obstruct and arterial vessel. A retrospective study using dexamethasone for lumbar transforaminal epidurals 2 months follow-up showed good relief comparable to particulate steroids [ 43 ]. A study from Kennedy et al. showed comparable outcomes using dexamethasone vs triamcinolone, although more injections were needed in the dexamethasone group [ 44 ].
An inexperienced practitioner may not be well-versed in all safety considerations or standard anatomic references of needle placement as recommended by the International Spinal Intervention Society Guidelines [ 45 ]. In light of this data, the use of “universal precautions” is recommended in performing transforaminal injections at any level. This should include a thorough review of the MRI by the practitioner. If the patient has known peripheral vascular disease or significant risk factors, further consideration may be given to the use of nonparticulate steroids. Live injection of contrast dye should be used prior to injection of steroids to ensure there is no arterial flow. When available, digital subtraction angiography may be used to further confirm the absence of arterial flow, particularly when injecting at higher lumbar levels where the artery of Adamkewicz is more likely to be present.
Limitations of this study were that all patient's historical information was not available. Also, the images in this study are from presumed underlying vascular disease, which was the clinical basis of obtaining these studies, and may not represent the arterial anatomy in all patients. However, the spine is said to only have enough blood supply to meet its metabolic demands [ 46 ]. It would follow that vascular disease could make patients at higher risk of infarction due to a tenuous blood supply although the data to support this assumption does not exist.
Another limitation of this study is that the medullary arteries that communicate with the longitudinal arteries of spinal cord could not consistently be visualized. This is because a special CT sequence appropriate flow rate is required to obtain this information. A subsequent study of CT-angiograms dedicated to the lumbar spine is being planned to obtain this information.
The dataset did not included L5 or sacrum as this was a retrospective study using CTA of the abdomen and pelvis that were performed in patients with significant aortoiliac disease. The studies were not optimized with the proper contrast flow rate and volume to reliably opacify the L5 intervertebral arteries. An additional prospective study could be performed to observe whether our findings with flow rates adjusted so as to opacify these arteries.
In summary, this study illustrates that an artery is may be present in either the safe triangle or in Kambin's triangle and sometimes even both. Although recent papers have suggested the infraneural approach via Kambin's triangle may be safer than the subpedicular or supraneural approach via the safe triangle, there is still a possibility an artery will be present in Kambin's triangle and thus a risk of arterial injection still exists. This highlights the reason to use necessary precautions including reviewing available imaging and using live injection of contrast dye to observe for vascular flow. Adherence to the safety guidelines from the International Spine Intervention Society in performing these procedures should be practiced in all L-TFESIs [ 45 ].
The authors would like to acknowledge Mitchell Maltenfort for his statistical assistance.
Funding sources: No funding was necessary in this publication.
Disclosure: The authors have no disclosures.