Neuro -radiology
Neuroradiology (1988) 30:232-238
© Spfinger-Verlag 1988
CT-diskography in the evaluation of the postoperative lumbar spine Preliminary results N. Grenier 1, J.-M. Vital 2, J.-F. Greselle 1, O. Richard 1, B. Houang 1, H. Pinol-Daubisse 1, J. Senegas 2, and J.-M. Caille 1 Services de 1Neuroradiologie et 2 d'Orthop6die, Groupe Hospitalier Pellegrin-Tripode, Bordeaux, France
Summary.Thirty-three patients with recurrent sciatica after lumbar-disk surgery were studied for recurrent herniated nucleus pulposus (HNP) by CT alone and CT-diskography (CTD). Twenty-six patients underwent surgical reexploration allowing correlation with CTD. CTD made a correct diagnosis of recurrent HNP in twenty-one patients showing an extravasation of the contrast medium from the disk space into the medullary canal. In two cases CTD was positive for recurrent HNP but surgery showed only fibrosis. The amount of contrast leak was very small in these two cases along the posterior common longitudinal ligament, and the junction with the disk was very narrow. Because of their clinical presentation three patients with negative studies were operated upon. One showed only fibrosis and the two others showed an associated disk fragment. Among the twenty-one true-positive cases, seven showed a combination of recurrent HNP and scarfing. Contamination of CSF by the contrast medium through the dura was observed in two patients. Although a prospective comparative study between CTD and IV-contrast-enhanced CT is necessary, CTD appears to be a useful diagnostic procedure for recurrent HNP after surgery of the lumbar spine.
Key words: Disk herniation - CT - CT diskography - Postoperative changes
Recurrent leg pain after surgery of the herniated lumbar disk is a frequent and difficult problem involving 25 to 40 per cent of those operated upon [1, 2]. This pain may be related to various processes but the three most frequent are fibrosis, recurrent herni-
ated nucleus pulposus (HNP) and bony stenosis. Both clinicians and radiologists have to recognize the exact etiology to undertake the appropriate treatment. After surgery for a HNP, the crucial dilemma is to differentiate a recurrent HNP from postoperative scarfing or fibrosis. Only the former requires a surgical correction. A new operation in case of fibrosis will not correct the symptoms and can even make them worse, leading to further scar formation [1, 3]. Clinically, low back pain and radiculopathy in the same territory as that present before surgery can be encountered in both recurrent HNP and scarfing [1, 2]. Acute onset of symptoms is more likely to be related to HNP than to fibrosis which usually becomes progressively worse after surgery. However, such a clinical distinction is hazardous in most cases. Plain CT has been shown to be relatively inefficient for this differential diagnosis [1, 2, 4, 5] although a recent report showed that it could make a correct diagnosis in 60 per cent of the cases [6]. For most authors, intravenous (IV)-contrast-enhanced CT with high doses of contrast medium is now considered the most helpful technique [1, 3, 5, 7]. Fibrosis, however, is not always enhanced and small disk fragments may fail to be recognized [7]. In addition, some disk fragments may also enhance after IV contrast [6]. MR, especially with gadolinum infusion [8] and perhaps with gradient echoes [9], may be valuable for distinguishing HNP from scarring but further studies are still necessary for their evaluation. Similarities between dynamics of iodine and paramagnetic contrast media will probably lead to the same difficulties. Whereas CT-diskography was recently shown to be helpful in preoperative studies by a few authors [10, 15], its usefulness for postoperative spine exami-
233
nation was only suggested in one [15]. This led to our study in postsurgical symptomatic patients to distinguish between these two entities by using surgical correlation. Strict comparison with IV-contrast-enhanced CT was not achieved because this examination was not available for all the patients.
Material and methods
Time for fibrosis to develop in the spinal canal and for the posterior common longitudinal ligament (PCLL) to heal is unknown. We postulated that patients showing extravasation of contrast medium from the disk space into the medullary canal through a rent in the outer annulus and the PCLL, at least three to six months after surgery were positive for recurrent disk disease. On the other hand this examination was considered as negative for H N P and suggestive of scarring if the contrast medium did not extend into the epidural space. We tested our hypothesis on thirty-three patients who were studied by CT-diskography between May, 86, and March, 87. All presented recurrent symptoms after lumbar spine surgery. There were twenty-two men and eleven women with ages ranging between 26 and 67 years. Radiculopathy appeared in thirty-two subjects between 3 months and 9 years after surgery. The last patient was never relieved by the second surgical procedure. Sciatica occurred in the same territory as that before surgery in all cases: L5 in 22 cases and $1 in 11. Selection of patients for CT-diskography depended upon plain CT results. This procedure was indicated only if soft tissue density areas without evidence of scarring were seen in the epidural space and if the recurrent pain could not be explained either by a bony stenosis or by a H N P at another level. We excluded patients with CT showing classical signs of retraction on the thecal sac and considered as positive for fibrosis [1, 2]. All patients showed a soft tissue area in the epidura! space, before IV contrast. Only five IV-contrast-enhanced CT were performed: four of them showed an enhancement of this area which was observed about ten to fifteen minutes after infusion. Diskography was the first step of the procedure using a classical extradural posterolateral approach with fluoroscopic control. Puncture of the disk was achieved under local anesthesia without any sedation. An 18 G diskography needle was introduced close to the annulus. Then a 22 G Chiba-type needle was passed through the lumen of the first and advanced into the disk. Good positioning of the
needle tip was confirmed by two orthogonal views. Criteria for positioning are not as strict for an operated disk as for a non-operated one because injection in the first type allows diffusion of contrast medium into the whole degenerated space. Two to three milliliters of Iopamidol (Schering) were injected testifying to the degeneration of the disk which normally does not accept more than 1.5 ml. The end of injection was marked by an increase in resistance or production of similar pain to that presented before by the patient. A.P. and lateral views were obtained before withdrawing the needle. All CT examinations were performed one to three hours after diskography. During this time, the patient was told to walk a few minutes to increase intradiskal pressure and improve diffusion of contrast. A C G R 1200 (Thomson-C. G. R.) scanner was used with 5 mm contiguous angled slices and a high resolution 512 x 512 matrix for reconstruction. Five sections were usually necessary for exploring one disk space. Covering the entire interval between the pedicles of the two adjacent vertebrae is mandatory because migration of disk fragments is frequent in recurrence. One millimeter thin sections were added if necessary in the middle of the disk or on the abnormal area to avoid partial-volume effects. Images were photographed with both narrow windowing to evaluate non-opacified soft tissues and the thecal sac, and large windowing for studying diffusion of the highly concentrated contrast into the canal and bony structures. IV infusion was never added during this procedure, neither were coronal nor sagittal reconstructions performed. Twenty-six patients were reexplored surgically a few days after CTD at the previously operated level allowing us to correlate CT-diskography findings with peroperative findings. Table 1. CT-diskography results among 33 patients
Positive CT-diskography Negative CT-diskography Total
Non-operated patients
Operated patients
3
23
4
3
7
26
Table 2. Comparative results of CT-diskography and surgical findings (26 patients) Surgical finding
HNP
Scar
21 2
2 1
CT-diskography findings HNP Scar
234
Fig. l a - d . a Precontrast CT-scan at the L4-L5 level shows a soft-tissue mass with low density (49 UH) in the left anterolateral epidural space, b After IV infusion of contrast there is an enhancement of the density in the same area (93 UH). e Diskography shows a posterior extravasation of contrast into the canal (arrow). d On CT-diskography this area is contaminated by the extravasated contrast medium injected into the disk. A recurrent H N P was found by surgery Fig.2a-d. a CT without contrast at L4-L5 level shows an anterior area with soft-tissue density, b IV-enhanced CT shows a slight peripheral rim of enhancement (arrows). c On CT-diskography at the same level we note an extravasation of contrast with a central area of low density along the posterior aspect of L5. d At the disk level we recognize the posterior track of contrast through the annulus (arrows). A downward-migrated H N P was found peroperatively Fig.4a, b. True-negative study, a CT without contrast at the L4-L5 level. There is a bulging annulus with relatively narrow recesses. b On CT-diskography at the same level contrast outlines the posterior margin of the disk. A small track of contrast appeared peripherely (arrow) but did not extend into the epidural space. On surgery, only fibrosis was found Fig.Sa, b. False-negative study, a On the precontrast study at the L4-L5 level the disk appears degenerated with a vacuum phenomenon and bulging annulus. In the left anterolateral epidural space there is a nodular soft-tissue area impinging upon the thecal sac (arrow). b On CT-diskography, opacification of the disk is almost complete without any extravasation in the epidural space. A recurrent H N P was removed peroperatively Fig.6a, b. a CT without contrast at level L4-L5 reveals an obliteration of the anterior epidural fat. b After diskography there is an extravasation of contrast into this soft tissue (arrows) with a contamination of CSF (arrowhead). A recurrent H N P with a dural defect was found at surgery
235 Fig.3a-d. False-positive study, a CT without contrast at the IA-L5 level shows a protruding disk with a small impingement on the left anterolateral aspect of the thecal sac. b Soft-tissue area replaced the normal epidural fat in the right lateral recess of S1. e, d On CT-diskography there is a leak of contrast enhancing the soft-tissue area. Note the narrow track (arrows)through the PCLL. Fibrosis was found at surgery
236 Results
CT-diskography results are summarized in Table 1. Circumferential bulging of the disk was present in every patient because of postoperative degeneration. Twenty-six cases (78%) showed extravasation of contrast in the epidural space. Configuration of this extravasation was nodular in all cases (Figs. 1 and 2). Enhancement of the entire soft-tissue-density area was observed in thirteen cases and surrounded by a non-enhanced part in the rest of them. Among all these cases considered as positive for recurrence, twenty-three underwent surgery. The three remaining patients were not operated upon because of general contra-indications. Peroperative findings compared to CT-diskography findings are summarized in Table 2. Twenty-one were recurrent disk fragments. We observed two false-positive cases (Fig. 3) where the amount of extravasated contrast was very small compared to the true-positive cases. In one of these false-positive cases, a technical error with a first injection outside the disk was responsible for contrast present in the foramen. Among the true positive cases, disk material migrated downward or in the lateral recess of the underlying vertebrae in eight (Fig.2). None of them migrated upward. Among the four cases where IV-contrast-enhanced CT showed an enhancement of the epidural tissue, this enhanced area was, in two, the one occupied by the disk fragment opacified during diskography (Fig.l). Negative CT-diskography was present in seven patients. Four of them were not operated upon because of this criterion. However, three went to surgery because of their clinical presentations: all showed a circumferential bulge with a slight postedor asymmetry without any connection with the epidural soft-tissue areas. One was confirmed as being only fibrosis (Fig. 4). Two (67%) appeared to be associated with a nonopacified disk fragment (Fig. 5). In two patients with recurrent H N P we observed a contamination of the CSF by the contrast medium through the dura in spite of an extradural posterolateral approach for puncture of the disk. It appeared as an enhancement of the CSF density in front of the HNP (Fig. 6). A dural fissure was identified peroperatively in both cases.
Discussion
Recurrent leg pain after surgery of the lumbar HNP is a frequent clinical dilemma because of several possible etiologies: extradural fibrosis, central canal
stenosis from bone overgrowth, facet syndrome, focal arachnoiditis or diskitis and recurrence of HNP [1, 2]. After diskectomy, scarring occurs along the posterolateral surgical approach and in the anterolateral part of the epidural space replacing the normal fat. It appears in the weeks following surgery as a healing after hematoma formation [2]. The outer annulus and the PCLL which are disrupted are never surgically repaired. Healing occurs progressively by scar-tissue formation in the anterior epidural space. However, the time necessary for these structures to heal is unknown. It is probably related to the time for fibrosis to develop. If the disk has not been completely removed by surgery, new strains or trauma can produce a recurrence of disk-material migration through the defect in the PCLL. Fibrosis develops systematically after a surgical procedure and when a recurrent disk fragment occurs, it migrates into this scar. CT has been shown valuable in identifying secondary bony changes, deformities and retraction of the thecal sac by secondary scarfing, and replacement of the epidural fat by soft tissue [2, 61. IV injection of high doses of contrast can produce an enhancement of the periphery of the annulus [3, 6, 7], enhancement of fibrosis in 65 per cent of cases [6], and enhancement of the periphery of recurrent disk fragments [3, 5, 6]. However, some authors have emphasized that several millimeters in thickness of a protruding disk or the entire disk fragment could enhance, mostly if imaging was delayed after the infusion [6]. In our series two patients showing enhancement of the disk fragment after IV infusion showed extravasation of contrast with CT-diskography proven by surgery. Although a controversy has recently appeared in the literature about the actual usefulness of diskography in diagnosis of HNP [10, 15], most of the authors agree that this technique does not contribute much or is only exceptionally necessary for its assessment. None have discussed its potential value for recurrent HNP recognition. Diskography allows a direct visualization of HNP because the contrast medium is injected directly into the disk. The problem with CT is in identifying a diskal structure in the extensive fibrosis. We, therefore, tried to visualize directly these fragments with opacification by a diskal approach. We used CT coupled with diskography because CT is more sensitive to small contrast leakage in the epidural space, as has recently been shown [10]. This technique allows a better analysis of the posterior margin of the disk, a localisation of the disk fragment in the fibrosis and its eventual extension upward or downward. Although we did not try to
237
compare diagnostic capabilities of CT-diskography versus IV-contrast-enhanced CT, this former technique appears interesting for diagnosis of recurrence. Twenty-one cases of recurrent H N P were recognized. However, patients with a CT diagnosis of fibrosis and a negative CT-diskography were not operated upon. The regime probably produces an underestimation of recurrence. In our series, three patients with negative studies were surgically explored. Surgery found two unsuspected disk fragments. Two explanations for this phenomenon can be proposed: first, it might be due to an insufficient amount of contrast injected into the disk; second, the track in the PCLL might be obstructed after passage of the fragment by an increase in pressure on the epidural site. Excluded fragments can also be recognized by this technique (three surgically proven cases in our series). Further studies will, however, be necessary to understand why and in which situations some fragments can be missed. Posterior leak of contrast occurred in 26 of our cases (78%). Among the twenty-three patients with positive examinations who had operations, twentyone (91%) were true positive for HNP. The two false-positive cases showed only a very small leak of contrast. In the first case it formed an extravasation along the PCLL with a very narrow neck (less than 4 ram) corresponding to the small sequellar track through the outer annulus and the PCLL (Fig. 3). In the second the contrast extended slightly beyond the limit of the disk, but it was misinterpreted as positive in spite of its narrow shape. All the truepositive cases showed a connection with the annulus larger than 5 mm. Analysis of the junction between the contrast leak and the annulus is therefore mandatory, mainly with thin slices to avoid partialvolume effects. Although further studies remain necessary, diagnosis of recurrence is questionable when this junction appears narrow with a small amount of contrast in the canal or when the contrast does not extend beyond the disk border, when the disk border is visible. All true-positive cases showed a large contamination of the epidural space. They all showed an asymmetrical posterior shape. Residual non-opacifled soft tissue around the disk fragment was present in seven cases and was interpreted as associated fibrosis. Opacification of a disk fragment was in most cases inhomogeneous with a relatively less dense central area. This phenomenon is probably related either to the heterogenous distribution of contrast generally observed by diskography in the fissures of a degenerated disk or to a certain amount of contrast located in the epidural space surrounding the fragment. Migration of recurrent disk frag-
ments is not infrequent. CT-diskography is very helpful in demonstrating it. It occurred in eight cases, always downward, with a lateral situation in the recess of the underlying vertebrae in four. The proportion of surgically proven recurrent H N P is larger in our series than in any other report. In a total of 33 patients, surgery revealed 23 cases of recurrence (70%). Our high positive rate for recurrence can be explained by the fact that we excluded from CTD patients with evidence of fibrosis or normal studies with CT. However, we have to assume that CTD is very inaccurate in excluding recurrent HNP, as shown by our two false-negative cases among 3 operated upon (67%). CSF contamination by the injected contrast medium occurred in two patients. In these two cases the disk approach for puncture was extradural. This can only be explained by a dural breach by the H N P as sometimes occurs with transdural herniated disks. These preliminary results show that, by assuming the postulate of a probable healing of the outer annulus and PCLL with time, we can get a high accuracy for diagnosis of recurrent H N P by CTD. We decided to limit exploration by CTD to patients operated upon at least three months earlier because scarring is supposed to develop in weeks following surgery [2]. Ross et al. [16] in a recent study with M R showed healing of the outer annulus 2 to 6 months after diskectomy. The high rate of falsenegative studies, however, reveals an inability to exclude recurrence. The selection bias of our patients for surgery, which excluded negative studies, may even underestimate the recurrence rate. Comparative studies of IV-contrast CT, CTD, and M R with gradient echos and Gadolinium infusion appear now mandatory to set up a better assessment of symptomatic patients after surgery of the lumbar spine.
References 1. Teplick JG, Haskin ME (1983) Computed tomography of the postoperative lumbar spine. AJR 141 : 865-884 2. Mall JC, Kaiser JA, Heithoff KB (1983) Postoperative spine. In: Newton T, Potts DG (eds) Computed tomography of the spine and spinal cord. Clavadel Press, San Anselmo, pp 187-204 3. Teplick JG, Haskin ME (1984) Intravenous contrast-enhanced CT of the postoperative lumbar spine: improved identification of recurrent disk herniation, scar, arachnoiditis and diskitis. AJNR 5:373-383 4. Braun IF, Lin JP, Benjamin MV, Kricheff II (1984) Computed tomography of the asymptomatic postsurgical lumbar spine: analysis of the physiologic scar. AJR 142:149-152 5. Schubiger O, Valanis A (1982) CT differenciation between r e -
238 current disk hemiation and post-operative scar formation: the value of contrast enhancement. Neuroradiology 22: 251-254 6. Firooznia H, Kricheff II, Rafii M, et al. (1987) Lumbar spine after surgery: Examination with intravenous contrast-enhanced CT. Radiology 163: 221-226 7. Yang PJ, Seeger JF, Dzioba RB, et al. (1986) High-dose IV contrast in CT scanning of the postoperative lumbar spine. AJNR 7:703-707 8. Hueftle MG, Modic MT, Ross JS, (1987) Post-operative epidural fibrosis versus recurrent disk herniation in the lumbar spine: assessment with Gd-DTPA enhanced MR imaging. RSNA, Chicago, p 251 9. Koschorek F, Brinkmann G, Gremmel H (1987) Value of fast field echo imaging in recurrence of lumbar disk hemiations. RSNA, Chicago, p 251 10. McCutcheon ME, Thomson III WC (1986) CT scanning of lumbar diskography: a useful diagnostic adjunct. Spine 11: 257-259 11. Bemard TN (1987) Don't discard diskography (letter). Radiology 162:285
12. Shapiro R (1986) Current status of lumbar diskography (letter). Radiology 159:815 13. Scullin DR (1987) Lumbar diskography (letter). Radiology 162:284 14. Errieo TJ (1987) The role of diskography in the 1980s (letter). Radiology 162: 285-286 15. Sachs BL, Vanharanta H, Spivey MA, et al. (1987) Dallas discogram description. A new classification of CT/Discography in low-back disorders. Spine 12:287-294 16. Ross JS, Masaryk TJ, Modic MT, Bohlman H, Delamater R, Wilber G (1987) Lumbar spine: Postoperative assessment with surface-coil MR Imaging. Radiology 164:851-860 Received: I0 February 1988 Professor J. M. Caille Service de Neuroradiologie Groupe H6spitalier Pellegrin-Tripode Place Am61ie-Raba-L6on F-33076 Bordeaux Cedex France