Lumbar Spinal Stenosis: How Is It Classified?

Gregory D. Schroeder, MD; Mark F. Kurd, MD; Alexander R. Vaccaro, MD, PhD, MBA

Disclosures

J Am Acad Orthop Surg. 2016;24(12):843-852. 

In This Article

Classification of Lumbar Stenosis

No clinically applicable and validated classification of spinal stenosis has been published, and this has substantially limited the development of an evidence-based algorithm for treatment. Stenosis is commonly classified by descriptive elements, such as etiology (ie, degenerative or congenital), location (ie, central, lateral recess, foraminal, extraforaminal), and severity of the narrowing on advanced imaging (ie, mild, moderate, or severe).[9] The diagnosis of spinal stenosis is often associated with spondylolisthesis or scoliosis, and the presence of instability substantially alters the required surgical intervention. Further complicating matters is the fact that the definitions of these seemingly simplistic classifications are not uniform in the literature; in addition, the aforementioned classifications are based on imaging findings, and the severity of the stenosis on images does not always correlate with the severity of the patient's symptoms.

Congenital Stenosis

In 1976, Arnoldi et al[10] proposed the first modern classification for lumbar stenosis, dividing it primarily into congenital or acquired (ie, degenerative) stenosis. Congenital stenosis was defined as a bony dysplasia leading to a narrowing of the spinal canal, with further subdivision into idiopathic and achondroplastic congenital stenoses.[10] No specific values for defining congenital LSS have been adopted. Verbiest[11] recommended defining absolute congenital stenosis as a sagittal diameter of the bony canal <10 mm and relative congenital stenosis as a diameter between 10 and 12 mm; these absolute measurements have not been universally accepted, however, because they include only patients with central bony stenosis, not those with congenital stenosis secondary to a trefoil-shaped canal (ie, a canal with a narrow midsagittal diameter that results in notable lateral recess stenosis) or ligamentous derangements.[12]

In the largest study to date that attempted to characterize congenital stenosis, Kitab et al[13] prospectively compared 66 patients aged <50 years with persistent neurogenic claudication and congenital stenosis on MRI with 45 age-matched patients who had undergone MRI for a single episode of low back pain that lasted <10 days. They found that patients with congenital stenosis had multilevel stenosis from L3 through L5, but involvement of L1, L2, and S1 was rare. A statistically significant decrease in the ratio of the spinal canal cross-section area to the vertebral body cross-section area (P < 0.05) and a decrease in the ratio of the AP spinal canal diameter to the AP vertebral body diameter were identified at all lumbar levels in patients with congenital stenosis (Figure 1). In addition, decreases in the ratio of the transverse spinal canal diameter to the transverse vertebral body diameter and in the interlaminar angle were identified only at the stenotic levels[13] (Figure 1). The authors performed a qualitative analysis of axial MRI scans and identified three patterns of congenital stenosis: stenosis primarily caused by a decreased AP canal diameter with a flattened appearance and compressed canal contents (ie, short pedicles); stenosis primarily resulting from a decreased interlaminar angle (ie, flat lamina); and a global decrease in the size of the canal resulting from a decrease in the AP and transverse diameters along with a decrease in the interlaminar angles.[13]

Figure 1.

Axial T2-weighted magnetic resonance images demonstrating the cross-section areas of the spinal canal and the vertebral body (A); the cross-section areas of the thecal sac and the spinal canal (B); the laminar angle (C); and the AP and transverse diameters of the spinal canal and vertebral body (D). (Reproduced with permission from Kitab SA, Alsulaiman AM, Benzel EC: Anatomic radiological variations in developmental lumbar spinal stenosis: A prospective, control-matched comparative analysis. Spine 2014;14[5]:808–815.)

These findings are somewhat consistent with those reported by Singh et al,[14] who performed a smaller prospective, control-matched, radiographic analysis comparing 15 patients with symptomatic congenital stenosis (as identified by a senior surgeon) with 15 asymptomatic matched control subjects. Singh et al[14] reported that patients with congenital stenosis had significantly shorter pedicles at every lumbar level (P < 0.05) and significantly smaller AP diameter and cross-section area of the spinal canal (P < 0.05 for each); however, they did not find a difference in the transverse diameter of the spinal canal at any level, and they found only a substantial trend toward a decrease in the transverse width at L3 (ie, a width of 21.0 mm in patients with congenital stenosis versus a width of 24.3 mm in control subjects; P = 0.08). The reported differences between the two studies may be the result of type II error in the study by Singh et al[14] (their sample size was substantially smaller than that of Kitab et al[13]).

Although a detailed discussion of congenital stenosis in patients with achondroplasia and other developmental disorders is beyond the scope of this article, it is critical to remember that patients with achondroplasia often have symptomatic congenital spinal stenosis secondary to short pedicles and a thick lamina.[15] In addition, patients with achondroplasia are more likely to have stenosis caudally because the interpedicular distance decreases in the caudal aspect of the achondroplastic spine.[15]

Location of the Stenosis

Compression of the neural elements can occur either centrally or laterally, with lateral compression commonly subdivided into lateral recess, foraminal, and extraforaminal zones (Figure 2). The combination of broad-based disk-osteophyte complex and ligamentum flavum hypertrophy often leads to central stenosis. In contrast, facet hypertrophy and associated osteophytes commonly result in lateral recess stenosis. Foraminal stenosis can result from a substantial loss of disk height, foraminal disk protrusions or osteophytes, or angulation in the setting of degenerative scoliosis, whereas extraforaminal nerve root compression most commonly is caused by a far lateral disk herniation.

Figure 2.

Schematic demonstrating the different locations of spinal stenosis.

Radiographic Severity of Lumbar Stenosis

The radiographic criteria for lumbar stenosis are unclear. In a recent systematic review of the literature, Andreisek et al[16] reported that 14 parameters can be used to define LSS and that the definitions of these parameters vary substantially. In a separate article based on a literature review and a survey of experts, Andreisek et al[17] identified 27 possible radiologic criteria for lumbar stenosis. Perhaps the most commonly used classification of severity of spinal stenosis was proposed by Lurie et al.[9] They qualitatively defined mild stenosis as a decrease of less than one third in the space available for the neural elements. Moderate stenosis was defined as a decrease of one third to two thirds in the space available, and severe stenosis was a decrease of more than two thirds in the space available. To determine the interobserver and intraobserver reliability of this approach, three radiologists and one orthopaedic spine surgeon evaluated the MRI scans of 20 patients who participated in the spinal stenosis cohort of the SPORT. The reliability of the method varied by location of the stenosis, with the interobserver reliability being substantial for central stenosis (κ = 0.73), moderate for foraminal stenosis (κ = 0.58), and only fair for lateral recess stenosis (κ = 0.49).[9] Despite the fact that this method is commonly used in the literature, its reliability has not been independently validated. Another simplistic and often-used method for quantifying central stenosis is the cross-section area of the dural sac. When this area is <100 mm2, the stenosis is considered moderate; when it is <75 mm2, the stenosis is considered severe.[18,19]

Recently, more complex individual classifications have been proposed for central and foraminal stenoses. Schizas et al[20] proposed a classification of the severity of central stenosis based on the morphology of the dural sac on MRI (Table 1). To validate the classification, 57 intervertebral disk segments from 37 patients who underwent surgery were reviewed by two senior radiologists, two senior spine surgeons, and two junior orthopaedic physicians; the developers of the classification reported moderate interobserver reliability (κ = 0.44) and substantial intraobserver reliability (κ = 0.65).

Lee et al[21] offered a more simplistic MRI-based classification system for central stenosis. Stenosis that does not eliminate the cerebral spinal fluid (CSF) anterior to the cauda equina on axial T2-weighted images is grade 0; stenosis that results in no CSF anterior to the cauda equina but still allows for separation of the cauda equina is mild or grade 1; stenosis that leads to some cauda equina bunching is moderate or grade 2; and stenosis that leads to no space between the elements of the cauda equina is severe or grade 3 (Figure 3). To validate this classification, the developers reported on 61 MRI scans of the lumbar spine that were reviewed by four musculoskeletal radiologists. The interobserver reliability was substantial to excellent for all levels (κ = 0.73 to 0.95), and the intraobserver reliability was excellent (κ = 0.86 to 0.90).[21] Furthermore, Park et al[22] independently validated this system by having two radiologists review 160 MRI scans of the lumbar spine, and they reported substantial interobserver reliability (κ = 0.78). Additionally, the authors reported that no patients with grade 0 stenosis had neurologic symptoms, and almost all patients with grade 3 stenosis had symptoms. However, the clinical relevance of grade 1 and 2 stenoses was unclear.

Figure 3.

Schematic and axial T2-weighted magnetic resonance images demonstrating the grading scale proposed by Lee et al21 for central lumbar stenosis. A and B, Stenosis that does not eliminate the cerebral spinal fluid (CSF) anterior to the cauda equina on axial T2-weighted images is graded as zero. C and D, Stenosis that results in no CSF anterior to the cauda equina but still allows separation of the cauda equina is considered mild or grade 1. E and F, Stenosis that leads to some cauda equina bunching is consideredmoderate or grade 2. G and H, Stenosis that leads to no space between the elements of the cauda equina is considered severe or grade 3. (Adapted with permission fromLee GY, Lee JW, Choi HS, Oh KJ, Kang HS: A new grading system of lumbar central canal stenosis on MRI: An easy and reliable method. Skeletal Radiol 2011;40[8]:1033–1039.)

Foraminal stenosis is best identified on sagittal T1-weighted images and is often defined as a foraminal diameter of <3 mm or a foraminal height of <15 mm.[23] To better classify foraminal stenosis, Lee et al[24] recently proposed the use of four grades. Grade 0 indicates no foraminal stenosis; grade 1 indicates mild foraminal stenosis, which is defined as the loss of perineural fat in either the transverse or the vertical direction; grade 2 indicates moderate foraminal stenosis, which is defined as the loss of all perineural fat with no morphologic changes to the nerve root; and grade 3 indicates severe foraminal stenosis, which is defined as stenosis that results in a morphologic change to the nerve root. To validate this classification system, the developers had two radiologists evaluate 576 foramina in 96 patients, and they found excellent (κ ≥ 0.80) interobserver and intraobserver reliability at all levels. This classification is similar to an earlier system proposed by Wildermuth et al,[25] who similarly divided foraminal stenosis into four grades based largely on the presence of perineural fat detected on MRI (Table 2). Park et al[26] independently validated and compared the reliability of the two systems by having two radiologists classify foraminal stenosis in 91 consecutive patients who underwent MRI of the lumbar spine. They found similar interobserver reliability between the two classifications (κ = 0.77 for the system by Lee et al[24] and κ = 0.73 for the system by Wildermuth et al[25]).

Finally, although no objective radiographic parameters have been universally accepted for the diagnosis of lateral recess stenosis, Steuer et al[23] conducted a systematic review and found the following measurements were useful in identifying the condition: a lateral recess height of ≤2 mm (ie, defined as the narrowest distance between the superior articular facet and the posterior border of the vertebral body), a lateral recess depth of ≤3 mm (ie, defined as the distance between the superior articular facet and the posterior vertebral body at the cephalad border of the pedicle), and a lateral recess angle <30° (ie, defined as the angle between the lines parallel to the floor and the roof of the lateral recess; Figure 4).

Figure 4.

A, Axial T2-weighted magnetic resonance image demonstrating the lateral recess height—the narrowest distance between the superior articular facet and the posterior border of the vertebral body (line). B, Axial T2-weighted magnetic resonance image demonstrating the lateral recess depth—the distance between the superior articular facet and the posterior vertebral body at the cephalad border of the pedicle (line). C, Axial T2-weighted magnetic resonance image demonstrating the lateral recess angle—the angle between the lines parallel to the floor and the roof of the lateral recess (lines).

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....