J. D. Meyer1

R. E. Latchaw1

H. M. Roppolo1 K. Ghoshhajra2

Z . L. Deeb3

Received July 29, 1981; accepted after revi­sion November 30, 1981 .

Presented at the annual meeting of the Ameri­can Society of Neurorad iology, Chicago, April 1981.

, Department of Radiology, Division of Neuro­radiology, University of Pittsburgh School of Med­ic ine, Presbyterian-University Hospital, DeSoto at O'Hara Sts. , Pittsburgh, PA 1521 3. Address re­print requests to J. D. Meyer.

2 Department of Radiology, Allegheny Valley Hospital, Natrona Heights, PA 15084.

3 Department of Radiology, Allegheny General Hospital, Pittsburgh, PA 15213.

AJNR 3:223-228, May/ June 1982 0195-6108/ 82 / 0303-0223 $00.00 © American Roentgen Ray Society

223

Computed Tomography and Myelography of the Postoperative Lumbar Spine

Postoperative myelographic changes in the thecal sac, epidural tissues , and bony canal , as well as nerve roots, may be difficult to interpret. A series of 32 postoperative patients, all of whom had a metrizamide myelogram and subsequent lumbar computed tomogram, was reviewed to examine the ability of computed tomography to recognize abnormalities when the myelogram is equivocal or uninterpretable. Criteria to distin­guish recurrent herniated disk from postoperative changes are presented , including the demonstration of mass densities similar to and in continuity with the intervertebral disk. In 12 reoperated cases, five recurrent herniated disks and two new herniated disks were diagnosed and confirmed. In 20 nonreoperated cases , no recurrent her­niated disks were identified, although two new herniated disks were found at levels not believed clinically significant. Computed tomography after metrizamide myelography appears to be a reliable technique for distinguishing abnormalities in the postoperative spine.

Difficulties in myelographic exam ination of postoperative lumbar spine are well known. These are caused by previous oi l myelography and operation, often superimposed on developmental variation , degenerative changes, and the alter­ation by acute and chronic disk disease. In particular , recurrent herniated disk or a new disk herniation cannot be accu rate ly or cons istently recognized by mye­lography in the postoperative lumbar spine.

Previous reports [1 -6] have demonstrated the usefulness of high resolution computed tomographic (CT) scan ning of the lumbar spine with and without metrizamide in various conditions. It would seem that the method might be useful to clarify problems encountered in the postoperative patient as well. We retro­spectively studied 32 patients to examine this possibility.

Materials and Methods

Clinical

The study group comprised 32 men and women of average age 45. All had at least one previous myelogram (usually iophendylate) and lumbar su rgery, usually a laminectomy, for hern iated disk . Many had had several operations and myelograms. Backache and/or sciat ica either persisted or recurred in all pat ients and all had a fai lure of conservati ve treatment before a decision to restudy the spine. Twelve patients were reoperated after myelography and CT scann ing.

Technical

All patients were studied by metrizamide myelography, followed by a lumbar spine scan on the General Electri c 8800 Scanner . Metrizamide at 190 mg / ml concentration in a volume of 10-1 2 ml was used. Filming was performed immed iately in the semierect posit ion , generally without fle xion or extension views. Delayed fi lming was not performed . All CT scans were obtained within 4 hr, usually 1 hr, with the patient supine and the legs

224 MEYER ET AL. AJNR:3, May / June 1 982

straight. Most were scanned at L5-S1 and L4-5, with abou t half at L3-4. Adjacen t sections of 5 mm thickness were scanned from about ped ic le to pedic le through the d isk space. Gantry angle was chosen by localizing film as th e c losest ang le parallel to the disk at the respecti ve levels. No attempt was made to correct for lordosis by changing the patient' s position . Scans were obtained at 120 kVp, 600 mAo Routine image display at 2X magnification was used. No reformatt ing or sag ittal/ coronal reconstruct ion was used. A window width of 500 was generally used, with window levels of about 80-120.

CT Diagnos tic Criteria

Herniated disk (figs. 1 - 5). Disk material was considered her­niated if a smooth and rounded mass of the same relat ive density as d isk materi al was found in re lation to th e disk or in continuity

A 8 Fig . .-Recurrent herni ated disk, proved by surgery. A , Extradural right­

sided defect (arrowhead) , prominent roots crossing L4-5 disk space, site of previous operation. B, Scan at L4 -5 interspace. Herniated disk fragment represented as smooth , rounded mass of disk density in epidural space (arrowheads) accompanied by distorti on o f dural sac and nonfilling o f root sleeve .

A 8 c

wi th it. Measurement of ac tu al CT numbers was not performed since the relative (qualitative) density value was found to be reliable. (In contrast, scarring was noted to produce patterns of mixed density and even radiolucent mass [see figure 8 below). Herniated disk material was usually posterolateral in the epidural space and distorted the thecal sac (figures 1 B, 2C, and 5A) . Distortion that was near but not directly related to the thecal sac deformity pro­duced by the disk fragment was considered to be produced by scar ti ssue without herniated disk material (Fig . 2A). Root sleeves were noted to be unfilled (figs . 1 B, 2B, 2C, 3B, 4B, 5A) or distorted (figs. 3 B and 4B) in cases of demonstrable herniated disk . Root sleeve distortion, per se, was not reliable as a diagnostic cr iterion of herniated disk material alone because it was also produced by scarring (fig. 6B). Bony irregu larities cou ld be read ily distinguished from herniated disk mater ial (Figs . 3B and 11 B). Recurrent her­niated disk was distinguished from new hern iated disk that occurred

A 8 Fig . 3. -Recurrent herni ated disk proved by surgery. A, Poor filling of S1

root sleeves bilaterally. (arrowheads). Previous operation was at thi s level, L5-S1 . B, Same level. Rou nded mass of disk density in anterolateral epidural space (arrowheads) accompanied by non filling of left S 1 sleeve and poor filling of the right S1 sleeve (arrow). Prominent edge of vertebral body (asterisk) .

Fig . 2.-Herniated disk, proved by surgery. A, Generalized irregu lar asym­metry secondary to scarring, with prom­inent extradural defect at L5-S 1 on left (arrowheads). Nonspec ific root de fects at L4 - 5 and over L4 (arrow) . Site of previous operation was L4-5 . B, New herniated disk. CT scan at L5-S 1 . Smooth, rounded ext radural density (ar­row) with minor impression on opaque column, which appears narrowed sec­ondary to scarring . This was not a site of previous operation. C, Probable recur­rent herniated disk . Scan at L4-5. Very prominent " bulge" of disk at site of pre­vious operati on (black arrows). Lamino­tomy operat ion defect (white arrow).

AJNR:3 , May / June 1982 POSTOPERATIVE LUMBAR SPINE 225

at a level not previously explored (fig . 2). Scar (figs. 6-10) . In contrast to herniated disk material, scarring

secondary to operative intervention was represented by a replace­ment of the epidural ti ssue (u sually fat) by denser ti ssue. The density of the scar ti ssue was variable and often lucent (Figs. 88 and 98) . The scar tended to be a band of irregular shape in the anterior epidural space (Figs. 78 and 108) or a relatively lucent distortion of the thecal sac (figs. 28 and 98). Root sleeves were kinked or deformed in an angu lar manner also, but not di splaced by a mass (figs. 68, 98 , and 108). This lack of mass, except in fi gure 8 , was the main distinguishing characteri stic of scar ti ssue.

Stenosis (fig . 11). Stenoti c pattern s were divided into develop­mental, central, and lateral in this stud y. Developmental stenosis resulted from general o r focal canal narrowing due to short pedic les, thick laminae, and thick face t joints . The sagittal diameter was reduced. Central stenosis was considered to be acquired laminar thickening and facet enlargement or fu sion bone encroaching on

A B Fig . 4. -Recurrent herni ared disk , proved by surgery. A, Nonfilling of ri ght

S1 root sleeve (open arrow) and widening of left S1 rool sleeve (closed arrow). This is site of previous operation, LS- S1. B, Similar fin dings accom­panied by central herni ation (black arrows). Widened root sleeve (arrowhead). Laminotomy defect (white arrow).

Fig. S.-Recurrent herniated disk , proved by surgery. A, Poor S 1 root sleeve fillin g bilaterally (arrowheads) and asymmetry with poor filli ng at L4-S (ar­row). Previous operations were at L4- S and LS- S1. B, At LS- S1. Herni ated disk material (arrowheads) on right with non­fi lling of S 1 root sleeve but good filling of left S1 root sleeve. C, Norm al L4- S rool sleeves are well filled (arrowheads) and disk margin is intact.

A

the canal centrally . Th e lateral stenosis pattern refers to narrowing of th e peripheral portion of th e canal, lateral recess or root exit ca nal.

Indeterminate (fig. 12) . In a small group, the canal was sur­rounded by dense ti ssue with no visualiza tion of ep idural fa t or root sleeves. The density of the ti ssue was not lucen t and scar tissue could not be distinguished from d isk materi al.

Normal (figs. 5C and 7C) . In th e absence of any of the above criteria, the CT scan find ings were considered normal at the level in question.

Results

Myelography

All myelograms were abnormal at least at one leve l, usu­all y the level in question . At L3-4 , the myelog ram was

A B Fig. 6. - Scar, not proved by surgery. A, Oblique spot film . Sleeve widen­

ing at LS- S1 on left (arrowheads) , site of prior operal ion. B, Same level. Flattening of sleeve (arrowhead) but no mass (nol confirmed surgica lly) probably due to scarrin g.

B c

226 MEYER ET AL. AJNR:3, May / June 1982

A B

Fig. 8. - Scar, operat ively confirmed . A, Extensive bilateral scarring pat­tern (arrows) . Arrowhead indicates level B. Prior operations were at L4-S and LS-S1 . B, Posterior displacemen t of dural sac (arrowheads) by " lucent .. mass: right S1 sleeve distorted (arrow) . All scar tissue at operation .

usually normal. Myelographic defects ranged from small root sleeve pressure defects, such as splaying and widen­ing, to compl ete block . The patterns often consisted of unfilled root sleeves, with or without some minor irregu larity of the thecal sac. As expected, many patients had extensive " scarring " patterns on examination . The herniated disks tended to produce a single unfilled root sleeve, but the sign per se was not specific for herniated disks. In fact, many leve ls that demonstrated unfilled roots by myelography alone were noted to show filled roots by CT. Irreg ular distortion of the thecal sac was a common feature also, and appeared to have litt le specific predictive value. Examples of myelograms are grouped with CT findings (figs. 1-12).

c

A

Fig . 7. - Scarri ng at L4- S, normal at LS-S1, not proved by surgery. A, Diffuse narrowing and poor sleeve filling pre­sumably secondary to scarring at L4- S (arrowhead) and LS-S 1 (arrow), both sites of previous operation. B, Through L4-S. Only minimal thickening and irreg­ular anterior epidural space. C, Through LS- S 1. Essentially normal anterior epi­dural space with only some loss of fat. Posterior disk edge (long arrows). Root sleeves are well fi lled (short arrow). The­ca l sac narrowed relative to myelo­graph ic finding.

B Fig . 9. - Scar, not confi rmed by surgery. A, Bilateral poor root sleeve

fi lling from L3- 4 to LS- S1. Usual LS-S 1 root sleeve location (arrowheads). Previous operati on was at L4-S. B, Through LS- S 1. Lateral dural sac distortion but no real mass effect. Find ings were most likely due to scarring .

CT Scanning

Table 1 summarizes the CT findings in all 32 patients. Tables 2 and 3 divide them into reoperated and nonreoper­ated groups, respectively. The operative findings at the principal level are indicated in table 4. In 12 patients who were reoperated , five recurrent herniated disks were iden­tified and four were surgically confirmed. The fifth patient , who had a recurrent herniated disk, also had a new her­niated disk at an adjacent level , which was believed to account for his symptoms and confirmed surgically. Two other patients had new herniated disks that were also sur­gica ll y confirmed. Although the total number of patients is

AJNR :3 , May / June 1982 POSTOPERATIVE LUMBAR SPINE 227

A B Fig. 1 a .- Scar, not operatively confirmed . A, Dural sac distortion and

nonfilling of right S1 sleeve (arrowheads); previous operation at LS-S1 . B, Same level. Moderate anterior epidural tissue thickening and loss of fat density. Sleeve on right (arrow) is normal. Left S1 sleeve does not fill as we ll .

A B Fig. 11 .-Lateral recess stenosis , confirmed by surgery. A, Poor filling of

S1 sleeves bilaterally, worse on left (arrowheads). Site of previous operation was LS-S1 . B, Same level. " Spur" arises from superior articular process o f S 1 related to spur that arose from posterior and lateral aspect of vertebral body (arrows) . S1 sleeve (arrowhead) displaced posteriorly.

small, CT myelography appears to be quite specific and accurate in the assessment of the recurrent or new herniated disk, provided that criteria for identification are ad hered to stri ct ly.

CT also appears to be accurate in demonstrating normal­appearing disk in the presence of confusing myelographic deformities . Many normal levels were, in fact, discovered by CT. Given the number and range of myelographic abnor­malities, more difficulty was expected. This appears to be in part due to improved root sleeve visualization by CT scan­ning .

B Fig. 12.-lndeterminate, no operation. A, Irregular distortion of dural sac

and poor fi lling of sleeves at L 1, LS, and S 1. CT level in B (arrowhead) . Previous operation at this level, L4-S . B, Small dural sac surrounded by dense tissue with no distinguishing charac teristics may represent pattern of scarring but et io logy is unknown.

TABLE 1: CT Diagnosis: All Patients

No. Patients (n ~ 32) Diag nosis

LS-S1 L4-S L3-4

Recurrent disk herniation 5 2 0 New disk herniation 0 1 0 Scar 8 8 7 Developmental stenosis 0 2 1 Central stenosis 0 2 0 Lateral stenosis 3 2 0 Normal 13 12 8 Indeterminate 1 3 0

TABLE 2: CT Diagnosis: Reoperated Patients

No. Pat ients (n ~ 12) Diagnosis

LS- S1 L4-S L3 -4

Recurrent disk herniation 3 2 0 New disk herniation 2 0 0 Scar 2 Developmental stenosis 0 1 Central stenosis 0 1 0 Lateral stenosis 0 0 Normal 4 5 3 Indeterminate 0 0

The number of indeterminate scans was small , a total of four levels on ly. It appears that thi s pattern , which may reflect scarring essentially, is not frequently encountered, despite the fact that patients were referred from many different sources.

In the group not reoperated, no recurrent herniated disks were found by myelography or CT . One patient had a new herniated disk at L4-5 , but it was not believed c linicall y

228 MEYER ET AL. AJNR: 3 , May / June 1982

TABLE 3: CT Diagnosis: Patients not Reoperated

Diagnosis

Recurrent disk herniation New disk herniation Scar Developmental stenosis Central stenosis Lateral stenosis Normal Indeterminate

TABLE 4: Operative Findings

Diagnosis

Recurrent herniated disk New herniated di sk Scar Stenosis

L5- S1

0 0 5 0 0 2 9

CT

5 2 2 4

No. Patients (n = 20)

L4-5 L3- 4

0 0 1 0 7 6 1 0 1 0 2 0 7 5 2 0

No. Patien ts (n = 12)

Surgery

4 2 2 4

significant. Roughly the same proportion of abnormalities was found in both groups, operative and nonreoperative (tables 2 and 3).

Discussion

Myelographic abnormalities in the postoperative patient have consistently led to difficu lty in interpretation [7-9]. Postoperative changes compound those produced by de­velopmental and degenerative abnormalities. High-resolu­tion CT evaluation after metrizamide myelography appears to offer clarifying information . This pre liminary report sup­ports the contention that the recurrent or new herniated disk can be demonstrated by high-resolution CT after metriza­mide myelography and distinguished from scar alone . Pro­vided that criteria of herniated disks are strictly adhered to as described above, accuracy is high.

Adjacent sections 5 mm thick were most often used to obtain data, but occasionally 2 mm overlap of a 5 mm section was necessary to demonstrate the continuity of disk material to the disk itself. Thin sections in the order of 1.5 mm on the General Electri c unit were not used because of the constraints of time and image noise and were not be­lieved to be necessary to demonstrate the findings . CT scanning without metrizamide would not have been diag­nostic si nce herni ated disk material would have been difficult to distinguish from scar, but also secondary changes on the theca l sac and root sleeves would have been much more

difficult to delineate. It is of interest that scarring does not produce a mass of

disklike density in close re lation to the disk space. More experience is necessary to determine whether thi s poten­tially usefu l d istinction is valid.

Spinal stenosis in its various forms may be a component of many spine diagnostic problems, and may itself produce symptoms typical of herniated disk [10-13]. Methods prior to CT scanning do not appear to have recognized its impor­tance .

In conclusion, CT scanning after metrizamide myelogra­phy seems to be a valuable technique to differentiate path­ologic anatomy in the postoperative spine . Not on ly does it aid greatly in the diagnosis of the herniated disk, but it helps identify normal levels in light of confusing myelographic patterns.

REFERENCES

1. Coin CG, Ying-Sek C, Keranen V, Pennink M. Computed as­sisted myelography in disc disease. J Comput Assist Tomogr 1977; 1 : 398-404

2. Arii H, Takahashi M, Tamakawa Y, Suzuki M, Shindo M. Metrizamide spinal computed tomography following myelog­raphy. Comput Tomogr 1980;4 : 117 -125

3. Naidic h T , King D, Moran C, Sagel S. Computed tomography of the lumbar thecal sac. J Comput Assist Tomogr 1980;4: 37-41

4 . Haughton V, Syvertsen A, Williams A. Soft-tissue anatomy within the spinal canal as seen on computed tomography . Radiology 1980; 134: 649-655

5. Williams A, Haughton V, Syvertsen A. Computed tomography in the diagnosis of herniated nucleus pulposus . Radiology 1980; 13 5: 95-99

6. Carrera G, Williams A, Haughton V. Computed tomography in sciatica. Radiology 1980; 137: 433-437

7. Mosely I. The oil myelogram after operation for lumbar disc lesions. Clin Radio/1977 ;28 : 267 -276

8 . Picard L, Roland J, Blanchot P, David R, Montaut J, Pourel J. Scarring of the theca and the nerve roots as seen at radiculog­raphy . J Neuroradio/1977;4 : 29-48

9 . Castan P, Bourbotte G, Herail J, Maurel J. Follow-up and post­operative radioculography, a radio logical analysis of 640 pa­tients. J Neuroradio/1977 ;4 : 49-93

10. Paine K, Huang P. Lumbar disc syndrome. J Neurosurg 1972;37 : 75-82

11. Ciric I, Mikhael M , Tarkington J, Vick N. The lateral recess syndrome, a variant of spinal stenosis. J Neurosurg 1980;53: 433-443

12. Quencer R, Murtach F, Post J, Rosomoff H, Stokes N. Post­operative bony stenosis of the lumbar spinal canal: evaluation of 164 symptomatic pat ients with axial rad iography. AJR 1978;131 :1059-1064

13. Kirkaldy-Willis W, Mcivor G, eds. Symposium: spinal stenosis. Clin Orthop 1976; 115: 2-1 44