S PINAL METASTASES are common in systemic

Transcription

1 Comparison of Spinal Magnetic Resonance Imaging and Myelography in Cancer Patients By Curt Hagenau, William Grosh, Michael Currie, and Ronald G. Wiley Spinal involvement by systemic malignancy is common, and often leads to extradural compression of the spinal cord and/or nerve roots by metastases. Rapid, anatomically accurate diagnosis is essential to the successful management of these patients. We compared spinal magnetic resonance imaging (MRI) with conventional myelography in a series of 31 cancer patients being evaluated for myelopathy (N = 10), or back/radicular pain (N = 21). All patients were evaluated between April 1985 and July 1986, and underwent both studies within ten days of each other (median, two days). MRI was performed on a 0.5 Tesla Technicare unit with a body surface coil, and results compared with standard contrast myelography. All studies were reviewed separately and in a "blinded" fashion. MRI and myelography were comparable in S PINAL METASTASES are common in systemic malignancy, and frequently lead to epidural compression of the spinal cord or nerve roots.1, 2 A common sequence produced by a spinal metastasis consists of local back or neck pain, followed by the development of radicular pain, then myelopathy, with motor, then sensory, and finally sphincter dysfunction." Early, accurate detection and characterization of the anatomy of these lesions must be made if the patient is to be spared irreversible paraplegia and incontinence. It must be kept in mind that benign pathologic processes of the spine can also affect these patients as well; disc herniations 4 and cervical spondylosis, for instance, can mimic malignant root or cord compression. Diagnostic imaging of spinal lesions in cancer patients has relied on myelography for many years. 5 Magnetic resonance imaging (MRI) recently has been proposed as an alternative, 6 perhaps even a better choice.' In this setting, MRI has several potential advantages over myelography, including noninvasiveness, greater patient tolerance, and the ability to image bony structures and paravertebral soft tissue as well as neural structures. This study was undertaken in an effort to identify which of these two modalities is currently superior in the evaluation of cancer patients with suspected epidural metastases. detecting large lesions that produced complete subarachnoid block (five of ten patients with myelopathy, three of twenty-one patients with back/radicular pain). In 19 of 31 patients, smaller but clinically significant extradural lesions were found. In nine of 19 cases, these lesions were demonstrated equally well by both modalities; in nine of 19 cases, these lesions were demonstrated by myelography alone; in one of 19, a lesion was demonstrated alone. Given our current technology, myelography appeared superior to MRI as a single imaging modality. However, MRI may be an alternative in patients where total myelography is technically impossible or unusually hazardous. J Clin Oncol 5: by American Society of Clinical Oncology. METHODS One hundred seven patients were evaluated for suspected epidural metastases between April 1985 and July 1986 at Vanderbilt University Hospital, Nashville, TN and affiliated hospitals. During this period, an effort was made to obtain simultaneous myelography and spinal MRI studies to enhance diagnosis. Thirty-five patients actually received simultaneous studies of corresponding regions of the spine, while 27 patients received only MRI, and 45 patients received only myelograms. Four of the 35 patients who received simultaneous studies were excluded from the study because the films were unavailable. Studies were considered "simultaneous" if performed within one to ten days (mean, two days) of each other without significant intervening therapy or clinical change. MRI was performed on a Technicare 0.5 Tesla superconductive magnet at Vanderbilt University Hospital. Standard scanning protocol included sagittal images of the appropriate area using spin echo techniques of TE = 38 ms and TR = 500 ms, TE = 45 ms and TR = 2,000 ms, and TE = 90 ms and TR = 2,000 ms. If areas of pathology were identified on the initial sagittal studies, or if there was clinical suspicion of a lesion at a specific level, then transverse images through that area were sometimes obtained (depending on whether the study was performed before or after a systems software update). From the Department of Neurology, Department ofmedicine, Division of Oncology, and the Department of Radiology, Vanderbilt University Medical Center, Nashville, TN. Submitted January 2, 1987; accepted April 29, Address reprint requests to Ronald Wiley, MD, PhD, Department of Neurology, Vanderbilt University Medical Center, 2100 Pierce Ave, Nashville, TN C 1987 by American Society of Clinical Oncology X/87/ $3.00/0 Journal of Clinical Oncology, Vol 5, No 10 (October), 1987: pp

2 1664 Table 1. Patient Characteristics Median age, 62 years (range, 11-86) Sex, 16 males, 15 females No. of Primary Neoplasm Patients Case No. Breast 7 2, 6, 7, 13, 19, 23, 27 Prostate 5 3, 4, 14, 21, 35 Lymphoma 5 8, 18, 24, 25, 26 Lung 3 1, 16, 20 Myeloma 2 15, 17 Melanoma 1 9 Fallopian tube carcinoma 1 22 Renal cell carcinoma 1 33 Chondrosarcoma 1 5 Liposarcoma 1 28 Chronic lymphocytic leukemia 1 31 Cervix 1 32 Adenocarcinoma of unknown primary 1 30 Small-cell cancer of unknown primary 1 29 Total 31 Myelograms were performed using a standard, cut film myelography system with either water-soluble or oily contrast media. The choice of contrast material depended on the status of the patient and preference of the radiologist. Myelograms and MRI scans were reviewed in a "blinded" fashion by one investigator (M.C.), who knew only the patient's age and primary malignancy. MRIs were read first, then at a HAGENAU ET AL later date all of the myelograms were read, so that knowledge of one study would not readily bias interpretation of the other. Charts were reviewed for all 31 patients. In comparing the relative merits of the two diagnostic tests for each individual case, the clinical evidence was used to establish the probable diagnosis. No patients died acutely and surgery is rarely performed at our institution for epidural metastases in patients with known cancer. Consequently, tissue confirmation of the radiological diagnoses was not available. The comparison reported in the present study reflects the test the managing physicians used to guide their decision making. RESULTS Table 1 describes the characteristics of the study patients. The patients consisted of two groups: those whose symptoms and signs suggested a myelopathy (group A, n = 10), and those evaluated because of back pain, or pain, weakness, and/or numbness in a radicular or nonspecific pattern but with no evidence of myelopathy on exam (group B, n = 21). The results of the comparison of myelography and MRI in these two patient groups is presented in Tables 2 and 3. Table 4 summarizes the lesions found, and Table 5 compares MRI and myelography in demonstrating the lesions. The term "cord compression" is used to mean complete or > 80% block of contrast on myelography, or complete obliteration of the subarachnoid space on MRI. "Other extradural defects" Table 2. Group A: Patients With Myelopathy at Presentation. Comparison of MRI and Myelogram Results Cord Compression Other Extradural Defects Case Defect Modality Best Defect Modality Best Level(s) and No. Found Defining Defect Level(s) Found Defining Defect Description 2 Yes Equal T 12 Yes Myelo EDT, SRC L3.4 not seen 25 Yes MRI (on myelo could not T 2, L1 No get dye below T 2 ) 33 Yes MRI (on myelo could not T 7, L1 No get dye between blocks) 35 Yes Equal T, No 24 Yes Myelo, cervical spondylo- C3-6 No sis not seen 17 No Yes Myelo SRC at C 6 (degenerative disease), EDT T 12 -L 4 poorly seen 27 No Yes Equal SRC T8 seen only by MRI, EDT L4 seen only by myelo 9 No No 19 No No 22 No No Abbreviations: EDT, extradural tumor; SRC, symptomatic root compression; Myelo, myelogram.

3 Table 3. Group B: Patients With Back/Radicular Pain at Presentation. Comparison of MRI and Myelogram Results Cord Compression Other Extradural Defects Case Defect Modality Best Defect Modality Best Level(s) and No. Found Defining Defect Level(s) Found Defining Defect Description 4 Yes Myelo (MRI suboptimal, T7-8 Yes Myelo EDT C 2 shown best by showed greatest stenosis Ti 1.) myelo; SRC C 5, C 6 shown only by myelo 16 Yes Equal Cr-T, No 30 Yes Equal T No (Conus) 5 No Yes Equal EDT on right at T No Yes Equal EDT on left at C 6 with SRC C 6 18 No Yes Equal EDT anteriorly at T5-6 with SRC Ts, 6 20 No Yes Equal SRC left L 5 by tumor and bulging disc 23 No Yes Equal Mechanical compression fracture of T, with anterior indentation of spinal canal 26 No Yes Equal Symptomatic anterior extradural defect at C 4 _, due to degenerative disease 28 No Yes Equal EDT with SRC bilaterally at L4 32 No Yes Equal SRC L,, S, 3 No Yes Myelo EDT at S, seen by both; Myelo showed SRC left S, (present clinically). MRI showed cauda equina compression (not present clinically or on myelo) 7 No Yes Myelo EDT at T 1, L3.4 not seen 8 No Yes Myelo Symptomatic infiltration of S 1 roots, not seen 13 No Yes Myelo Both showed SRC left Ls; myelo showed EDT L45 14 No Yes Myelo Multiple cervical SRC due to degenerative dissease, seen only by myelo 15 No Yes Myelo Both showed anterior EDT at L1, only myelo showed upper cauda equina compression (present clinically) with a complete block 31 No Yes MRI Both showed S2 fracture, SRC bilateral L5 due to degenerating disc seen only 1 No No 21 No No 29 No No Abbreviations: EDT, extradural tumor; SRC, symptomatic root compression; Myelo, myelogram; MRI, magnetic resonance imaging. 1665

4 1666 Table 4. Summary of Lesions Found No. of Patients With a Lesion Group A Group B Total Category of Lesion (n= 10) (n=21) (n=31) Lesions causing cord compression Malignant 4* 3* 7 Spondylosis Noncompressive lesions (small extradural tumors or symptomatic root compression) Total with lesions *One patient also had a noncompressive lesion at a different level. refers to small extradural tumor masses that are not causing cord compression, or to symptomatic root compression. Tumor nodules on spinal roots were not seen by either modality. The following two cases are exemplary. Patient no. 28, a 33-year-old woman, underwent amputation of her left lower extremity because of a fibrosarcoma that had developed in her calf. Inguinal nodes were histologically positive for tumor. In the month following surgery, she developed progressive pain in her low back, radiating to the lateral aspect of her right leg. Examination revealed weakness of dorsiflexion on the right, but no sensory, sphincter, or reflex abnormalities. Percussion tenderness was noted over L 4-5. Her imaging studies are shown in Fig 1. MRI and myelopathy were felt to be equivalent in their demonstration of the epidural disease. Patient no. 13, a 50-year-old woman, was found 3 years previously to have infiltrating ductal breast carcinoma. On admission, she de- Table 5. MRI and Myelography Compared for their Ability to Identify and Define Extradural Defects No. of Patients Feature of Extradural Defect MRI Myelo Being Compared (%) (%) Total Cord compression: identified 7 (88)* 8 (100) 8 Cord compression: adequately defined 6 (75)* 6 (75)t 8 Other extradural defects identified 10 (53) 18 (94) 19 *One patient with cervical spondylosis was not identified, and in one case the level of greatest stenosis was inadequately demonstrated. tin two patients, the myelographer was unable to get dye both above and below the lesion(s). HAGENAU ET AL scribed a 2-month history of progressive pain in her low back and left thigh. Examination showed lumbar spine tenderness; weakness of the left psoas, glutei, and quadriceps; and numbness over the lateral left thigh. Her imaging studies are shown in Fig 2. Myelography was superior to MRI in demonstrating the epidural metastasis. DISCUSSION This comparison, although based on a limited number of synchronous studies, suggests that large epidural tumor masses compressing the cord are identified equally well by either MRI or myelography. However MRI missed one cord compression due to cervical spondylosis, and two of the patients with cord compression also had epidural tumor at other levels that was missed. Consequently, MRI proved superior only in two instances where the myelographer, despite lumbar and cervical punctures, was unable to get dye above and below the compressive lesion(s). Small epidural masses, and root compressions are better detected by myelography. Of the patients with small epidural tumor masses or symptomatic root compressions (asymptomatic defects due to degenerative disease not included), 47% (nine of 19) had lesions detected by myelography and not MRI. In 5% (one of 19) MRI showed a symptomatic root compression not detected by myelography. The detection of these smaller epidural masses and root compressions was often clinically important for the planning of radiation ports, and occasionally a surgical procedure. In many cases, MRI demonstrated nondestructive metastases, limited to the marrow of vertebral bodies, which were not detected by plain films or myelography. Also in a few cases, MRI demonstrated paravertebral masses not seen by plain films or myelography. However, identification of these entities was not included in our comparison of the two modalities, as they did not have direct impact on the diagnosis and management of epidural disease. The distribution of primary tumor types represented in our study population is similar to that of other series evaluating epidural metastases.8'9 The relatively small number of complete blocks found (eight of 31) probably reflects patient selection resulting from our policy of promptly

5 MRI V MYELOGRAPHY IN CANCER PATIENTS 1667 Fig 1. Metastatic fibrosoaroma of the left leg- MRI and myelogram equivalent. (A), Sagittal, Ti weighted MRI of lumbar spine. Posterior to the fourth lumbar vertebral body is a soft tissue mass, which narrows the thecal sac (thin arrow). The spinal nerves (curved arrow) are not compressed in this single projection. The L4 and L5 vertebral bodies exhibit some decreased signal intensity, suggesting early metastatic involvement (arrowheads). (B) and (C), Lumbar myelogram. There is an anterior impression on the thecal sac at the L4 vertebral body level (straight arrows). Note that the posterior cortex of L4 is intact (open arrow). The fifth lumbar nerve roots are obliterated at the level of the L4 pedicles, bilaterally (curved arrows), but the remaining spinal nerves are unaffected. Also note that the L4 and L5 vertebral bodies are normal (arrowheads). evaluating cancer patients with back and/or radicular pain. This policy is based on evidence that outcome of therapy of epidural spinal cord compression is directly related to neurologic status at the time of diagnosis. 8 s' 1 '" Many patients were evaluated initially with MRI or myelography alone, and when cord compression was found, they were treated urgently without obtaining a companion study. Many reports have described the usefulness of MRI in imaging diseases of the spine including spinal epidural metastases,'2-1 4 but few have directly compared MRI and myelography. Masaryk et al described four patients with cervical

6 1668 HAGENAU ET AL Fig 2. Metastatic breast cancer-myelogram superior to MRI,. (A), Sagittal, T1 weighted image of the lower lumbar spine exhibits some epidural fat and soft tissue anterior to the canal (arrows). However, no suggestion of nerve root abnormality was present. (B) and (C), Myelogram exhibits lateral and anterior compression of the thecal sac with compression of the left L5 nerve root (arrows). The thecal sac is displaced posteriorly from the vertebral body and the sac is narrowed in its lateral width. epidural neoplasms." The lesions were detected in all four patients by both modalities, but were felt to be better characterized in three of four patients. Yu et al reported 16 cancer patients evaluated with spinal MRI because of back pain. 6 Epidural metastases were seen in all 16, and two patients had epidural compression that had not been seen on earlier myelography. These two procedures have similar problems. Both myelography and MRI can be uncomfort-

7 MRI V MYELOGRAPHY IN CANCER PATIENTS able for the cancer patient with spinal metastases. Both take 45 to 90 minutes; are performed on a hard surface; and require that the patient lie still, although this is more often a problem with MRI. Being confined in the narrow tube of the MRI scanner induces claustrophobia in some patients. Other disadvantages of MRI are its higher cost and limited availability, particularly in emergencies. A disadvantage of myelography is its invasiveness, entailing the risks of dural puncture and contrast toxicity. In particular, as many as 14% of patients with spinal cord compression may acutely deteriorate neurologically after lumbar myelography,'6 and those with coagulopathies may experience local bleeding complications of the procedure." Neither of these complications were seen in the present series, and both have been rare in our institution over the past several years. This is offset to some degree by the value of obtaining cerebrospinal fluid for cytology, which is frequently needed in these patients. One additional advantage of myelography is that it provides the option of following with contrast-enhanced spinal computed to mography to further characterize levels of abnormality. 1 " CONCLUSIONS With our current technology, myelography remains the preferred initial imaging modality for the evaluation of suspected epidural spinal metastases. It is more sensitive than MRI for detecting small epidural metastases and root compressions, and provides cerebrospinal fluid for cytology at the same time. If the myelographer encounters difficulty getting dye both above and below a level of block, or between two tandem lesions, MRI would serve as a complementary second study. There are a few unusual situations were MRI may be preferred as a first choice-if the patient refuses myelography, or if myelography is contraindicated because of severe thrombocytopenia, neutropenia, or intracranial hypertension. MRI technology is advancing rapidly. Prospective studies to compare these two modalities will be needed in the future to provide rational guidelines for physicians striving to manage these patients in the most efficient manner. 1. Rodriguez M, Dinapoli RP: Spinal cord compression with special reference to metastatic epidural tumors. Mayo Clin Proc 55: , Posner JB: Neurologic complications of systemic cancer. Med Clin North Am 55: , Constans JP: Spinal metastases with neurological manifestations. J Neurosurg 59: , Carr BI, Goodkin R: Breast cancer with osseous metastasis and herniated lumbar disc-a cautionary tale. Cancer 56: , Bricout PB, Modur RS, Feldman MI: Importance of myelography in early diagnosis of spinal epidural disease. J Natl Med Assoc 73: , Yu E, Sarpel SC, Sarpel G, et al: Early diagnosis of spinal epidural metastasis by magnetic resonance imaging. Proc Am Soc Clin Oncol 5:10, 1986 (abstr 38) 7. Franklin EA, Berbaum KS, Dunn V, et al: Impact of MR imaging on clinical diagnosis and management: A prospective study. Radiology 161: , Gilbert RW, Kim JH, Posner JB: Epidural spinal cord compression from metastatic tumor: Diagnosis and treatment. Ann Neurol 3:40-51, Rodichok LD, Harper GR, Ruckdeschel JC, et al: Early diagnosis of spinal epidural metastases. Am J Med 70: , 1981 REFERENCES 10. Greenberg H, Kim JH, Posner JB: Epidural spinal cord compression from metastatic tumor: Results with a new treatment protocol. Ann Neurol 8: , Findlay GFG: Adverse effects of the management of malignant spinal cord compression. J Neurol Neurosurg Psychiatr 47:76-78, Han JS, Kaufman B, El Yousef SJ, et al: NMR imaging of the spine. AIR 141: , Aichner F, Poewe W, Rogalsky W, et al: Magnetic resonance imaging in the diagnosis of spinal cord diseases. J Neurol Neurosurg Psychiatr 48: , Rossi DR, Charney AS: Magnetic resonance imaging of the spine. Semin Neurol 6:84-93, Masaryk TJ, Modic MT, Geisinger MA, et al: Cervical myelopathy: A comparison of magnetic resonance and myelography. J Comput Tomogr 10: , Hollis PH, Malis LI, Zappulla RA: Neurological deterioration after lumbar puncture below complete spinal subarachnoid block. J Neurosurg 64: , Marton KI, Gean AD: The spinal tap: A new look at an old test. Ann Intern Med 104: , O'Rourke T, George CB, Redmond J, et al: Spinal computed tomography and computed tomographic metrizamide myelography in the early diagnosis of metastatic disease. J Clin Oncol 4: , 1986