Indications for Surgery

Spinal stenosis is the most common reason for lumbar spine surgery in adults older than the age of 65 years.¹ Lumbar stenosis occurs secondary to spondylotic changes at the facet joints, instability, or a congenitally small canal. Pathologic changes affect the central vertebral canal, lateral recess, and neural foramina (Fig. 64–1). Stenosis at the lateral recess from spondylotic changes is typically from the superior articular process of the caudal vertebra compressing the traversing nerve root, leading to neurogenic claudication (i.e., stenosis at the L4-5 level leading to compression of the L5 nerve root by the superior articular process of L5) (Fig. 64–2). The natural course of spinal stenosis is that a substantial proportion of patients do not automatically deteriorate and will remain unchanged or even improved by nonoperative treatment.² Ultimately, patient desire combined with failure of conservative treatment with physical therapy, activity modification, medication, and steroid injections drives the decision for operative treatment. Proper patient selection is critical to achieving a good outcome with spinal stenosis surgery. The ideal patient has symptoms of neurogenic claudication, which includes pain, numbness, and paresthesias in the posterolateral legs and thighs associated with prolonged walking or with activities causing back extension such as walking up stairs. Neurogenic claudication may also manifest as cramping and exhaustion of the lower extremities. Activities such as sitting, leaning forward on a walker or shopping cart, and riding a bicycle typically alleviate the pain. Lastly, lower extremity symptoms secondary to vascular claudication must be ruled out.

FIGURE 64–1 Lumbar stenosis is anatomically categorized into compression within the central canal, lateral recess (subarticular), and neural foramina.

FIGURE 64–2 Stenosis at the lateral recess leads to compression of the traversing nerve root by the superior articular process of the caudal vertebra (i.e., the L4-5 level leads to compression of the L5 nerve root as it traverses the lateral recess before exiting the foramen).

Deen and colleagues³ found that the most common cause of early failure after lumbar laminectomy was the absence of classical symptoms of neurogenic claudication coupled with the absence of stenosis on preoperative imaging studies. A predominance of low back pain over radicular pain is also associated with worse surgical outcomes.^4,5 In addition, a selective nerve root block is a good prognostic indicator of surgical outcome because patients who obtain greater than 50% relief of leg pain for at least 1 week after an injection tend to have greater than 50% relief of leg pain after surgery.⁶ Patients who do not respond to a selective nerve root injection and have pain for more than 1 year will generally have a poor surgical outcome.⁶ Surgical decompression can be done on an elective basis unless the patient has a rapidly progressing neurologic deficit and bowel/bladder dysfunction, both of which are rare presentations of spinal stenosis. Before surgical intervention we routinely obtain standing anteroposterior (AP), lateral, and flexion/extension radiographs, as well as magnetic resonance imaging (MRI) or computed tomography (CT) myelogram. It is important to recognize that instability is a dynamic process and may not necessarily be apparent on static supine MRI. Instability must be carefully assessed on flexion/extension radiographs in all patients. Any degree of instability is assessed, and all stenotic levels are identified. Although several surgical techniques have been described to treat lumbar stenosis, there is no clear evidence of the most effective technique.⁷ Surgical options include decompressive laminectomy with or without fusion, laminotomy, minimally invasive decompression, or placement of an interspinous device.

Laminectomy

In the absence of instability, laminectomy remains the gold standard for treating central, lateral recess, and foraminal stenosis. After induction of anesthesia, the patient is placed in the kneeling position on an Andrews frame (Orthopedic Systems, Union City, Calif.). Use of the Andrews frame has been shown to decrease vena caval and central venous pressures, leading to a reduction in blood loss when compared with a Cloward surgical saddle.^8,9 In addition, flexion of the hips reduces normal lumbar lordosis and widens the interlaminar space, making it easier to access the spinal canal. Use of a Wilson frame is also able to reduce lordosis from the lumbar spine during the decompression, although not to the same degree as the Andrews frame. It is important to note that placing a patient in this position can underestimate the true degree of stenosis when compared with the lordotic position obtained with a Jackson table. If a fusion is planned in addition to a laminectomy, we use a Jackson table in order to assist fusion in a lordotic position. If a Wilson frame is used in the fully cranked position for the laminectomy, it is important to uncrank the frame before the final fusion so as to prevent fusing in a nonlordotic position.

We believe that flexion of the knees with the use of pillows under the anterior calf is also important because it has the potential to reduce tension on the sciatic nerve. We routinely use 2.5× or 3.5× loupe magnification and an operating headlight. The relationship of the iliac crests to the lower lumbar levels on preoperative radiographs should be carefully examined to help guide the incision. A standard midline skin incision is made over the desired levels. The incision should be long enough to allow for exposure of the pedicles of the cephalad and caudal levels to be decompressed. For example, for an L3 to L5 pedicle-to-pedicle decompression, there should be enough exposure so that the inferior aspect of the L3 pedicle and the superior aspect of the L5 pedicle are easily palpated at the end of the decompression. Dissection is carried down to the thoracolumbar fascia with electrocautery. The fascia should be clearly exposed approximately 1 cm off of midline with a Cobb elevator to ensure a distinct layer for fascial closure at the conclusion of the case. The spinous processes are easily palpated and can be used to fine-tune the fascial incision for exposure of the appropriate levels. Electrocautery is then used to dissect just lateral to the spinous process, taking care to preserve the supra and interspinous ligaments. At this point a Kocher clamp is placed on the cephalad aspect of one of the spinous processes so that the clamp is in line with the pedicle of interest. Alternatively, a Woodson elevator can be placed in the intralaminar space to mark the appropriate surgical level. An intraoperative radiograph is then taken. It includes the Kocher clamp or Woodson elevator and the sacrum in order to confirm the appropriate level. This step is critical because 71% of wrong-level spine surgery occurs in the lumbar spine.¹⁰ Once the correct levels are identified, subperiosteal dissection of the paraspinal muscles is carried laterally with a combination of a Cobb elevator and electrocautery exposing the spinous process and lamina. During the dissection, the midlateral pars must be clearly identified so that it can be later used as a guideline for bony resection (Fig. 64–3). One should remember that the facet joints are superficial to the level of the lamina and midlateral pars and that as the dissection is carried out laterally, care should be taken to preserve the capsule of the facet joint. It has been demonstrated in cadaveric models that excision of the capsule and cartilage of the facets results in increased motion in both sagittal and axial planes, potentially leading to clinical instability.¹¹ For an isolated laminectomy, dissection should be carried out just to the lateral aspect of the facet capsule. Hemostasis with bipolar electrocautery is critical for visualization, particularly around the facet joint secondary to bleeding from the parafacetal arteries medial and lateral to the facets.¹² If a posterolateral fusion is planned, the dissection needs to extend laterally to the transverse process taking care to maintain the integrity of the intertransverse membrane between each level. The cephalad and caudal limits of the laminectomy are delineated with a Leksell rongeur. This includes the inferior half of the spinous process at the top of the decompression and the superior half of the spinous process of the inferior level to be decompressed. This is typically adequate to be able to palpate the pedicles of the cephalad and caudal levels to be decompressed. A Horsley bone cutter is used to remove the remaining spinous processes up until the spinous process/lamina junction (Fig. 64–4). The rongeur is once again used to thin the lamina. Any bleeding from cancellous bone should be controlled with bone wax. The decompression should begin centrally because the central zone is the last region to become stenotic. Because of the relatively shorter length, the oblique orientation, and larger attachment areas of the ligamentum flavum, the laminas of the lower spine have a much greater proportion of their anterior surface covered by the deep layer of the ligamentum flavum (Fig. 64–5).¹³ A curette is used to dissect the underlying ligamentum flavum from the inferior aspect of the lamina (Fig. 64–6). A Kerrison punch is then used to remove bone from the inferior aspect of the lamina using the ligamentum flavum as a protective layer. The decompression is carried cephalad until the ligamentum flavum ends and epidural fat or dura is encountered. In our experience this is approximately at the inferior aspect of the pedicle of the lamina being removed. In an L3 to L5 decompression, we recommend beginning at the inferior aspect of L4 and, once the dura is identified, moving on to removing the inferior aspect of L3 (Fig. 64–7A). This is followed by placement of a cottonoid between the dura and remaining laminar bridge of L4 and removing the residual L4 lamina and superior laminar edge of L5 (Fig. 64–7B). Adhesions between the dura and the flavum or lamina can be released with the use of a Woodson elevator or a Penfield No. 3 dissector. A central trough is thus safely created with complete removal of the lamina and flavum (Fig. 64–7C). The next step involves decompression of the lateral recesses. It is critical to adequately decompress the lateral recesses because it is well known that the most common technical error for early failure after lumbar laminectomy is inadequate neural decompression.³ A high-speed bur may be used to thin the lamina down and delineate the bony resection with extreme caution, particularly in the cephalad aspect of the lamina, where the cortical bone is thin and not protected by the ligamentum flavum. Lateral recess decompression should be performed with a Kerrison punch with care taken to preserve at least 50% of the facet joint (Fig. 64–8). Care must also be taken to preserve approximately 1 cm of the pars interarticularis in order to prevent an iatrogenic pars fracture. The location of the medial aspect of the pedicle to the midlateral pars (MLP) varies with each lumbar level. Su and colleagues¹⁴ recently described the percentage of pedicle medial and lateral to the MLP and found that with more cephalad levels, there is less pars when the decompression is carried out to the medial aspect of the pedicle (Fig. 64–9). Particular care should be taken at levels cephalad to L4 so as to not take too much of the lamina, leaving a narrow pars susceptible to a pars fracture. As such, with a decompression from L2 to S1, the laminectomy should appear to be trapezoidal with a narrower laminectomy trough in the more cephalad levels and a wider trough in the more caudal levels. Depending on the surgeon’s experience, the medial aspect of the inferior facet can be removed using an osteotome, exposing the underlying superior facet to be removed with a Kerrison punch. Cadaveric studies have shown that medial facetectomy does not affect lumbar spinal stability and, conversely, total facetectomy, even created unilaterally, makes the lumbar spine unstable.¹⁵ In addition, postlaminectomy CT scans of patients with resection of more than one half of the bone immediately above the flare of the inferior articular process at the level of the laminectomy resulted in a facet fracture.¹⁶ Undercutting the superior facet using a Kerrison punch assists bone removal while preserving the facet joint and the pars interarticularis. Because of the angle of approach, this maneuver should be performed from the opposite side of the operating table to ensure complete visualization of the dura during the decompression. Decompression should progress laterally until the medial wall of the pedicle is easily palpated with a Woodson or angled dural elevator. In the cranial caudal dimension, the pedicles of the superior and inferior segments should be palpable. Decompression of the lateral recess often leads to bleeding from the epidural venous plexus. Hemostasis is obtained with the use of bipolar electrocautery and hemostatic agents such as an absorbable gelatin sponge (Gelfoam) and cross-linked gelatin granules and topical human thrombin (Floseal). The last stage of the decompression involves foraminotomies at each level. Each nerve root should be traced as it passes underneath the pedicle and into the foramen. Any bone or soft tissue is removed with a Kerrison punch placed dorsal and parallel to the root as it exits the foramen. In the presence of exit-zone stenosis, bony spurs from the dorsal and lateral aspect of the superior facet should be resected. As in decompression of the lateral recesses, this should also be done from the contralateral side to prevent iatrogenic durotomy of the nerve root sleeve. A Woodson elevator or Murphy probe is used to assess the decompression by passing it into the foramen dorsal to the nerve root; a 4-mm ball tip probe should pass without difficulty.¹⁷ The nerve root can also be assessed for mobility by gently retracting the dural sac and nerve root medially into the canal with a Penfield No. 4 dissector; if adequately decompressed, the nerve should demonstrate 1 cm of medial displacement.¹⁷ At the end of the decompression, any sharp spikes of bone should be smoothed with a curette or a Kerrison punch to prevent it from lacerating the dura. Lastly, each level should be examined for an extruded herniated disc fragment, which should then be removed in a routine manner. We do not recommend a discectomy for well-contained discs. Before closure, hemostasis is once again obtained and the wound is irrigated. We routinely place a No. 15 French fully fluted Blake drain in the deep wound. The fascia is closed with interrupted figure-of-eight No. 1 absorbable sutures. The subcutaneous tissue is closed with buried 2-0 absorbable sutures, and the skin is closed with a nonlocked running 3-0 nylon suture or 3-0 absorbable subcuticular suture.

FIGURE 64–3 Exposure for a laminectomy must include clear delineation of the midlateral pars (MLP) at each laminectomy level as it provides a guideline for the degree of final bony resection. Dissection should be taken out to the lateral aspect of the facet, taking care to preserve the facet capsule.

FIGURE 64–4 The decompression begins by delineating the extent of bony resection with a rongeur. For an L3-5 decompression, this includes the inferior half of the L3 spinous process and the superior half of the L5 spinous process (shaded area). A Horsley bone cutter is used to remove the spinous processes up to the spinous process/lamina junction, which includes the entire L4 lamina.

FIGURE 64–5 A, In general, the ligamentum flavum originates from halfway up the lamina of the cephalad level and inserts onto the superior aspect of the caudal lamina. B, From a saggital view, the laminae of the lower spine have a much greater proportion of their anterior surface covered by the deep layer of the ligamentum flavum.

(From Olszewski AD, Yaszemski MJ, White AA III: The anatomy of the human lumbar ligamentum flavum. New observations and their surgical importance. Spine 21:2307-2312, 1996.)

FIGURE 64–6 A and B, A curette is used under the inferior aspect of the lamina to separate the ligamentum flavum from the overlying lamina.

FIGURE 64–7 A, In an L3 to L5 decompression, we recommend beginning at the inferior aspect of L4 and, once the dura is identified, moving on to removing the inferior aspect of L3. B, A cottonoid is placed between the dura and remaining laminar bridge of L4 so that the residual L4 lamina can be removed. C, The final central trough is created by removing the superior aspect of the L5 lamina.

FIGURE 64–8 A, Lateral recess decompression with a Kerrison punch preserving at least 50% of the facet joint (axial view). B, Final decompression after undercutting the lateral recess.

FIGURE 64–9 With more cephalad levels, there is less distance from the medial pedicle to the midlateral pars. Care should be taken at levels cephalad to L4 to avoid taking too much of the lamina, leaving a narrow pars susceptible to a pars fracture. With an L2-S1 decompression, the laminectomy trough should be trapezoidal with a narrower width at the cephalad levels.

(From Su BW, Kim PD, Cha TD, et al: An anatomical study of the mid-lateral pars relative to the pedicle footprint in the lower lumbar spine. Spine [Phila Pa 1976] 34:1355-1362, 2009.)

Arthrodesis After Laminectomy

Some authors have recommended that an arthrodesis be performed when stenosis is associated with instability or in the setting of a spondylotic spine associated with complaints of low back pain.²⁶ Grob and colleagues²⁷ prospectively evaluated the results of decompression with and without arthrodesis in 45 patients with stenosis without instability. Instability was defined as greater than 5 mm of motion in the saggital plane between segments or greater than 5 mm of lateral offset in the coronal plane.²⁷ All patients had significant clinical improvement compared with preoperative values at an average of 28-month follow-up with no significant differences regardless of whether or not a fusion was performed.²⁷ The results indicate that in the absence of instability, decompression with care taken to not destabilize the spine does not require an arthrodesis.²⁷ However, other authors in prospective observational studies have reported that in patients with stenosis (the majority of which did not have instability or scoliosis), decompression and arthrodesis versus laminectomy alone led to greater relief of back pain at 2-year follow-up.²⁸ The most likely explanation is that in patients with stenosis without instability, spondylotic changes contribute to a component of low back pain. We currently do not fuse patients who have stenosis secondary to spondylosis unless there is a significant component of low back pain associated with it.

Herkowitz and colleagues²⁹ reported on the results of 50 patients who had spinal stenosis associated with degenerative spondylolisthesis to determine if noninstrumented arthrodesis provided better results than decompressive laminectomy alone. At a mean of 3-year follow-up, patients with an arthrodesis had significantly more relief of pain in the back and lower limbs.²⁹ In regards to use of instrumentation for fusion, Fischgrund and colleagues³⁰ conducted a prospective study in 76 patients with spondylolisthesis and stenosis randomized to decompression and fusion with and without instrumentation. Although use of instrumentation increased the percentage of patients with a successful arthrodesis (82% vs. 45%), clinical outcomes at 2-year follow-up were similar between the two groups with approximately 80% excellent or good relief of leg and back pain.³⁰ However, longer-term follow-up (average 7 years, 8 months) of the cohort of patients who had an uninstrumented fusion in the study by Herkowitz and colleagues²⁹ and Fischgrund and colleagues³⁰ found that clinical outcome was excellent to good in 86% of patients with a solid arthrodesis versus 56% of patients with a pseudarthrosis (P = 0.01).³¹ Benefits of a successful arthrodesis over pseudarthrosis were demonstrated, contradicting previous reports with shorter-term follow-up that indicated no differences in clinical outcome between the two groups.³¹ Although use of instrumentation increases rates of fusion³⁰ and success of uninstrumented arthrodesis leads to better clinical outcomes,³¹ caution must be taken when implying that use of instrumentation leads to better clinical outcomes.

It is our current recommendation that arthrodesis with pedicular fixation be performed in the setting of a decompressive laminectomy if there is associated instability at the involved motion segments (>5 mm), degenerative scoliosis (curve progression or >30 degrees), revision decompression at the same level, or resection of greater than 50% of the facet joints.¹²

Surgical costs for patients with stenosis in the SPORT study were estimated by Tosteson and colleagues³² using the 2004 Medicare payment rate. The reported cost per Quality Adjusted Life Years (QALY) gained with surgical treatment of spinal stenosis was $77,600, and the reported cost per QALY gained with surgical treatment of stenosis with degenerative spondylolisthesis was $115,600.³² The majority of degenerative spondylolisthesis cases were treated with fusion, which carries a higher upfront surgical cost.³² With longer-term follow-up, the value of the surgical treatment of degenerative spondylolisthesis may improve, assuming that the observed improvements in quality of measures are lasting and that the need for additional treatment and revision surgery is minimal.³²

Laminotomy

In patients with primarily lateral recess stenosis, laminotomy is an alternative to laminectomy. Laminotomy involves performing the decompression through a microdiscectomy-like approach targeting the stenotic levels either unilaterally or bilaterally. Advocates argue that by leaving the midline structures intact, there is less chance of iatrogenic instability and back pain.³³ Other surgeons think little additional stability is gained by performing multiple laminotomies compared with a facet-sparing laminectomy because the majority of stability is provided by the intervertebral disc and the facet joint–capsule complex.³³ Thomas and colleagues³⁴ evaluated postoperative radiographs and MRIs in patients who underwent laminectomy or laminotomy. They found that a laminotomy can adequately decompress lumbar canal stenosis and that laminectomy and laminotomy have the same degree of postoperative listhesis.³⁴ Aryanpur and colleagues³⁵ described the results of decompressive laminotomies and foraminotomies in 32 patients with focal lateral recess stenosis. At last follow-up, 90% of the patients treated reported an excellent outcome with no significant postoperative morbidity or mortality, concluding that in a subgroup of patients with lumbar stenosis, multilevel laminotomies may be an acceptable alternative to laminectomy.³⁵

McCulloch described the results of a “less invasive” one- or two-level decompression and uninstrumented fusion in 21 patients with degenerative spondylolisthesis and stenosis.³⁶ The decompression is performed through bilateral fascial incisions with preservation of the midline structures and with the aid of a microscope.³⁶ This technique begins with excision of the lower half of the anterior portion of the cephalad lamina proximal to the origin of the ligamentum flavum. The insertion of the ligamentum flavum is then removed by excising the superior edge of the trailing (caudal) lamina. After removing the medial edge of the facet joint so that it is flush with the medial border of the pedicle, the stenosing ligamentum flavum is removed from the top down or from the bottom up (Fig. 64–10). Twenty of 21 patients had relief of their claudication and the overall fusion rate was 86%.³⁶ Costa and colleagues recently reported on the clinical and radiographic results of unilateral laminotomy and bilateral microdecompression of 374 patients with stenosis and found that 88% had a clinical benefit at a mean follow-up for 30 months.³⁷