Evaluation and Treatment of Tandem Spinal Stenosis

Joseph F. Baker, MCh, FRCSI


J Am Acad Orthop Surg. 2020;28(6):229-239. 

In This Article


Spinal stenosis can be considered in a variety of ways as follows: the presence or absence of symptoms; radiographic findings; or the underlying pathologic process(es) such as spondylosis, ossification of the posterior longitudinal ligament (OPLL), and ossification of the ligamentum flavum (OLF). Variations in the definition of spinal stenosis and, therefore, varied methodology in epidemiologic research predispose the literature to inconsistent findings when attempting to establish prevalence of TSS. Indeed, a number of epidemiologic approaches have been used in assessing TSS: these include population-based, patient cohort, and cadaver studies. Whether the stenosis is symptomatic or asymptomatic is another inconsistent feature of these studies, and a variety of grading systems have been used to define severity of radiologic disease.

Patient Cohort Studies

A small number of studies have analyzed selected patient cohorts to determine the prevalence of noncontiguous stenosis (Table 1). One must consider the demographic composition of these studies and whether the results apply to one's own clinical setting. A number of reports are from Asia where conditions such as OPLL and OLF are more common than in other geographic regions.

One of the largest studies to date assessed 931 subjects who had undergone whole-spine magnetic resonance imaging (MRI) as part of the Wakayama Spine Study in Japan, a large-scale population-based study. In this study, Nagata et al[4] excluded patients who had previous cervical or lumbar surgery and those younger than 40 years. They found a prevalence of imaging-based CSS of 25%, lumbar spinal stenosis (LSS) of 30%, and TSS of 11%. A higher prevalence of TSS was seen among those with congenital canal stenosis. Among those with TSS evident on MRI, symptomatic CSS and LSS were present in 10% and 19%, respectively.

In 319 patients treated for cervical spondylotic myelopathy (CSM), Kong et al[5] reported 12.3% presented with symptomatic LSS during follow-up. The main risk factor for the subsequent presentation of symptomatic LSS was increasing severity of CSS. The presence of congenital stenosis in the cervical spine appears as a risk factor for concomitant LSS—in a cohort of 80 patients with CSM, Miyazaki et al[6] demonstrated with CT myelography a greater prevalence of thoracic and lumbar stenosis in those with a congenitally narrow spine. Izuka et al[7] determined the prevalence of CSM in a cohort of 237 with symptomatic LSS to be almost 9%. After logistic regression analysis, only the Torg-Pavlov ratio was a predictor of CSM in patients with LSS—a value of 0.78 was predictive.

Although these studies are from Asian cohorts, some salient points may be taken. The prevalence of symptomatic TSS was approximately 10%. A recurring theme was that of a narrower lumbar canal predictably being associated with a narrower cervical canal. A reduced Torg-Pavlov ratio may be useful therefore in suggesting an increased risk of concomitant stenosis.[8]

Cadaver Studies

A series of studies using specimens from the Hamann-Todd Collection have provided additional details on the dimensions of the various regions of the spinal column. Bajwa et al[9] studied 1,072 specimens measuring pedicle lengths and canal dimensions in the thoracic and lumbar spine with digital calipers—the mean pedicle lengths were 5 and 5.5 mm, and the mean canal areas were 164 and 190 mm2 in stenotic and normal specimens, respectively—mean values across the entire study are not provided though. Defining stenosis simply as anteroposterior canal dimensions more than two SDs less than the mean, they reported lumbar stenosis in 57 and thoracic stenosis in 92. They failed to show a strong relationship between lumbar and thoracic stenosis—indeed, concomitant stenosis was found in only 1.4% of specimens. Furthermore, those specimens with a greater number of stenotic levels were at no greater risk of having concomitant stenosis elsewhere. The same authors also reported on the relationship between cervical and thoracic and cervical and lumbar noting, consistent with other reports, that there was evidence of association between TSS and a congenitally narrow canal.[10,11]

Assessing 440 specimens, Lee et al[12] reported that, based on raw data, 95 (22%) had at least one level of cervical stenosis and 74 (17%) at least one level of lumbar stenosis, both defined as a midcanal diameter of <12 mm. The presence of LSS resulted in a positive predictive value of 32%. The presence of CSS gave a positive predictive value of 25%. When applying correction for the size of the contemporary human skeleton, they determined the prevalence dropped to 5% and 6%, respectively, with only 1% having stenosis in both regions.

It must be acknowledged, however, that although useful to provide large-scale data and relationships between the different regions of the spinal column, cadaver studies of osseous structures alone do not consider the soft tissues, primary disk, and ligament that contribute to acquired stenosis seen in clinical practice and so likely underestimate the prevalence of stenosis seen in the clinical setting.