Chapter 174 Postoperative Imaging
The management of chronic degenerative spinal conditions in the United States is estimated to cost nearly $85 billion annually.1 Spine surgery to treat degenerative disease represents a large portion of this expenditure, and procedures such as lumbar fusion have increased dramatically over the last 20 years.2 In the past, routine fusion procedures for common indications such as spondylolisthesis did not necessarily require postoperative imaging, especially with the widespread use of fluoroscopy for intraoperative placement of instrumentation. Imaging was reserved for those situations in which preoperative symptoms did not resolve or new symptoms arose postoperatively. However, the margin for error in instrumentation position is now much narrower, despite a lack of evidence of neurologic injury. In addition, determining the correct level of surgery is paramount, and when anatomic abnormalities such as transitional vertebral segments raise doubt about the precise level of pathology, imaging is increasingly utilized. Increasingly, 3D intraoperative imaging, with real-time navigation, is being used to confirm the correct level and instrumentation position.
Normal Postoperative Spine
Analysis of the postoperative image requires understanding not only the anatomy but also the surgical approach, the goals of surgery, and any potential complications. Imaging abnormalities in the spine related to surgical trauma persist typically through the first 6 to 8 weeks. The extent of bony removal at surgery greatly influences the postoperative imaging appearance. The removal of bone may allow dorsal expansion of the dura through defects, which can resemble a pseudomeningocele, seroma, or postoperative hematoma. Additionally, asymmetry in the paraspinal muscle-fat planes related to surgical dissection is a normal finding. Edema may obscure muscular margins. MRI T2-weighted images, particularly fast spin-echo sequences, provide excellent resolution of water content, allowing visualization of edema or fluid collections.3
When the intervertebral disc has been violated or partially excised, distinction between epidural tissue edema and the disrupted anulus fibrosus and recurrent disc herniation or residual disc material can be complex.4 Well-vascularized granulation tissue usually enhances homogeneously with gadolinium administration, whereas disc material typically enhances poorly or, at best, heterogeneously. Nonenhanced MRI is approximately 85% accurate in achieving this distinction,5 whereas enhanced MRI is 95% to 100% accurate.6 Both fibrosus and disc can be hyperintense relative to bone on T2-weighted images. An important technical point for these evaluations is the timing of imaging following contrast administration. If imaging is delayed more than several minutes beyond the administration of the contrast medium, gadolinium seeps across the disc, which is then seen to enhance more homogeneously. To avoid this phenomenon and to increase the specificity for detecting fibrosus, imaging should commence within 2 minutes of gadolinium injection.
Epidural fibrosus is present in all patients following spine surgery to some degree. In some patients, formation of epidural granulation tissue is excessive and has been reported to contribute to residual pain and radiculopathy after surgery.7,8 However, other reports have contested this claim because granulation tissue is not thought to exert a compressive force.9 In the setting of recent discectomy and residual postoperative pain, enhancement of the epidural fibrosus, as well as the end plates and anulus fibrosus, may be easily mistaken for early signs of infection (Fig. 174-1).
Van Goethem et al. prospectively studied 34 patients with excellent outcomes following lumbar discectomy and obtained MRIs at 6 weeks and 6 months.10 Imaging findings in these asymptomatic patients mimicked diagnostic findings in patients with complications. Twenty percent of patients had recurrent disc herniation, and an additional 20% demonstrated nerve root enhancement at 6 weeks. At 6 months, contrast enhancement was seen along the surgical tract in all patients. Likewise, facet joint enhancement was also seen bilaterally and was attributed to mechanical stress during laminectomy.
A similar study reported MRI findings on 15 patients who had undergone anterior cervical discectomy and fusion (ACDF) immediately postoperatively, at 6 weeks, and at 6 months.11 Foraminal narrowing persisted in 66% of the first postoperative scans and did not resolve in the follow-up scans up to 6 months, despite symptomatic improvement. Nearly all cases demonstrated persistent edema, with high T2 signal, in the operative disc space at 6 weeks. In addition, all cases showed enhancement in this disc space at 6 weeks, and 50% persisted at 6 months.
Unresolved or Recurrent Symptoms
Recurrent Disc Herniation
The incidence of recurrent disc herniation in the immediate postoperative period is unknown. In addition, for the reasons discussed earlier, small fragments of retained disc are difficult to identify on imaging. The appearances of recurrent or residual disc are low signal intensity on T1-weighted MRI without enhancement after the administration of IV gadolinium in contrast with epidural fibrosus that typically does enhance. A large sequestered disc may have central high signal intensity on T2-weighted images (Fig. 174-2). In contrast to scar, disc fragments tend to have a smooth margin.
Approximately 10% of discectomy patients experience recurrent lumbar disc herniation requiring revision discectomy. The mean time for this complication in one study was 10.5 months.12 Patients with a larger preoperative anular defect and with a smaller percentage of disc volume removed had a significantly greater chance of recurrent disc herniation. Common hemostatic agents used in lumbar discectomy, such as oxidized cellulose (Surgicel), have been reported to cause immediate postoperative radiculopathy.13 The increasing use of minimally invasive techniques, with less direct visualization of the disc and nerve roots, may eventually result in a higher incidence of retained disc fragments. However, at present, the literature suggests that the rate of complications is the same for minimally invasive techniques as for traditional open surgery.14
Postoperative Hematoma
One of the most critical immediate postoperative complications to look for after spine surgery is a compressive hematoma, usually in an extradural location (Fig. 174-3). Fortunately, new onset of a major neurologic deficit in the lumbar spine is rare, due to canal space and mobility of nerve roots. However, in the cervical and thoracic spine, neural compression by a postoperative hematoma can have devastating neurologic consequences. Of nearly 12,000 adult spine operations performed over 10 years, the incidence of a major neurologic deficit immediately after surgery was 0.178%; in the cervical spine 0.293%, thoracic spine 0.488%, and lumbosacral spine 0.0745%.15 Epidural hematoma accounted for 38% of these complications, and it was the most common cause of immediate neurologic deficit. Other reasons for new deficit included inadequate decompression, presumed vascular compromise, graft or cage dislodgement, and presumed surgical trauma.
When a compressive hematoma is suspected, T2-weighted MRI is most sensitive for the detection of blood products in the spine.4 Instrumentation caused artifact on MRI and may obscure a subtle compressive lesion. It is sometimes necessary to also perform fine-cut CT with sagittal and coronal reconstructions to visualize hardware and bony anatomy. However, choice of imaging modality depends on the clinical scenario, since prompt reexploration is required with severe neurologic deficit.
The chances of a successful outcome following spine surgery depend on the initial indications, but generally are greater than 90% for common procedures such as microdiscectomy. However, a significant number of patients do not clinically improve following surgery, with estimates varying from 10% to 30%.16 In these instances, careful imaging of the operative site and spine is warranted.17 Several etiologies of continued pain and their imaging characteristics are discussed in the following sections.
Postoperative Infection
Radiographic diagnosis of postoperative spine infection is complicated—the time course of normal postoperative findings such as anular enhancement also mirrors when the same imaging characteristics represent pathologic infection. Postoperative infection may vary in its presentation, including spondylodiscitis, superficial cellulites or infected fluid collection, and paravertebral or epidural abscess (Fig. 174-4). Postprocedural discitis is relatively uncommon, with an incidence of less than 1%, although some studies reported an incidence as high as 3% of patients.18,19 Early in the infection, radiographs are not sensitive at detecting spondylodiscitis. CT may demonstrate destructive lesions and erosions in the vertebral end plates, as well as collapse of disc height. MRI is the best imaging modality for diagnosing discitis, with sensitivity and specificity of 93% and 97%, respectively.20 MRI may be supplemented with radionuclide scintigraphy, which has been reported to improve the specificity of MRI.21 The reported sensitivity and specificity of gallium scanning in detecting postoperative discitis is 89% and 85%, respectively.22 The MRI findings of postoperative spondylodiscitis include4,18
• Vertebral end-plate or marrow changes (low signal on T1- and high signal on T2-weighted images)
• Enhancing bone marrow adjacent to disc
• Enhancing spinal canal tissue
• Enhancing paravertebral soft tissue mass: rim (abscess) or homogeneous (phlegmon) enhancement
Although some of the aforementioned signal changes can occur normally after surgery (including disc enhancement), extensive contiguous enhancement of the disc and adjacent marrow is more consistent with infection. However, severe Modic changes from degenerative disease and aseptic discitis may appear indistinguishable from infection on MRI.23 Laboratory evaluations including serum white blood cell count, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) are often more helpful in diagnosing infection than is imaging.24 Both ESR and CRP are nonspecific inflammatory markers and demonstrate a postoperative peak from the trauma of surgery. Nevertheless, a second rise or a persisting elevation of CRP levels after surgery has a sensitivity and specificity of 82% and 48%, respectively, for infectious complications.24 In addition, when following documented spine infection to ascertain clinical improvement after long-term antibiotic administration, ESR and CRP levels are crucial as measures of infection clearance and responsiveness to antibiotics. Administration of gadolinium in serial MRI has also been reported to assist in the conservative management of nonspecific spondylodiscitis.25
Although epidural abscess is a rare postoperative complication, the overall incidence of this condition has been increasing to 2 cases per 100,000 people annually.25 An infected epidural collection may either enhance homogeneously, if phlegmonous, or in a rim fashion, as with abscesses. An epidural collection may cause neurologic symptoms from mass effect or thrombosis of spinal draining veins, and in either case, emergent decompression is required. Recently, limited surgical approaches have been advocated to prevent the spread of infection to other anatomic structures such as vertebral bodies or discs.26
Postoperative Arachnoiditis
Chronic arachnoiditis is another potential cause of the failed back surgery syndrome (FBSS).27,28 The potential causes of arachnoiditis are protean, but not proven. The inflammatory response may be to blood, infection, trauma, contrast medium, or any other intrathecally injected substance. The incidence of arachnoiditis after spinal surgery is approximately 3%.29 Clinical findings of radiculopathy do not necessarily correlate with the severity of the arachnoiditis as appreciated on imaging.30
Myelography, with and without CT, and MRI (particularly T2-weighted fast-spin echo sequencing) both depict arachnoiditis with high accuracy, and classical findings on each correlate well with each other (Fig. 174-5).29 MRI can detect arachnoiditis with 92% sensitivity, 100% specificity, and 99% accuracy.31 There are three typical imaging patterns of arachnoiditis:32
1. Nerve roots are conglomerated in the center of the thecal sac, representing mild disease.
2. An “empty” thecal sac caused by adhesions of the nerve roots to the walls of the dura, representing moderate disease.
3. An intrathecal soft tissue mass with a broad dural base that may obstruct the cerebrospinal (CSF) pathway. The mass has intermediate signal intensity and may show varying degrees of enhancement.
Postoperative Radiculitis
Enhancement of nerve roots is a common finding in a normal postoperative spine but may also represent a source of ongoing symptoms. The enhancement is caused by disruption of the blood-nerve barrier, either after surgery or from chronic or severe compression by a herniated disc. Root enhancement can be seen in asymptomatic patients for 6 to 8 months postoperatively, after which it usually resolves.33 Thereafter, root enhancement correlates well with radiculopathy.34
A recent study correlated patient symptoms after lumbar discectomy with MRI findings.35 A total of 120 postoperative patients underwent MRI to evaluate their nerve roots for enhancement, thickening, and displacement. The incidence of nerve root enhancement was 65.7%, and this MRI finding was associated with symptoms in the offending nerve root distribution, with sensitivity of 91.7%, specificity of 73.2%, and a positive predictive value of 83.7%. When all three imaging findings were present (enhancement, thickening, and displacement), the positive predictive value increased to 94.1%.
