Early Detection and Treatment of Spinal Cord Compression

In this issue of ONCOLOGY, Dr. Ruckdeschel addresses a subject that, fortunately, is not very common, but unfortunately for those in whom the problem occurs, the outcomes are almost universally poor. The subject is probably one of the most dreaded complications of advanced cancer-malignant spinal cord compression. On a positive note, since Dr. Patchell's plenary session presentation at the 2003 American Society of Clinical Oncology Annual meeting,[1] interest in metastatic spinal cord compression has been renewed and there is hope that future patients with this problem will fare better. In order to improve results in these patients, a theoretical framework can be proposed based on the natural history of the process (Figure 1).[2] In fact, Dr. Ruckdeschel addresses a number of these issues in the management of spinal cord compression as a means of improving outcomes for these patients. His discussion includes the importance of early detection, timely work-up of patients suspected of having cord compression, and optimal treatment strategies based on the patient's presenting clinical and radiologic features. Early Diagnosis
Dr. Ruckdeschel and colleagues were one of the first groups to suggest that patients be identified early, before the onset of irreversible neurologic damage.[3,4] Unfortunately, very few groups changed their practice based on this group's work, as evidenced by the proportion of patients who were not able to walk on presentation, even 15 to 20 years later. For example, in the Princess Margaret Hospital experience of 914 episodes of cord compression (1990- 1996), only 54% of patients were able to walk (with or without assistance) on presentation. This is similar to the experience at other institutions.[5-7] Patients clearly do better if they are treated before significant neurologic deficits are present. If one pools the data from all the prospective radiotherapy studies reported, 94% of patients with early spinal cord compression who are treated with radiotherapy when they are able to ambulate maintain that ability posttreatment.[8] This is tremendously better than the 13% and 38% who regain the ability to walk if they present with paraplegia or paraparesis, respectively.[8] Many groups are trying to identify which patients are at highest risk for cord compression. Where Ruckdeschel differs from the others is on the cardinal symptom of pain. Two groups failed to confirm the presence of back pain as a predictor of clinical or subclinical cord compression. Talcott et al[9] performed a multivariate analysis of patient, radiographic, and neurologic factors of 342 computed tomography (CT) scans in 258 patients to predict which patients were at highest risk for malignant spinal cord compression. They identified six predictive risk factors for this complication including the inability to walk, increased deep tendon reflexes, compression fractures on radiographs of the spine, the presence of bone metastases, bone metastases present for more than 1 year, and age less than 60 years. Talcott et al concluded that patients with none of the five risk factors had a 4% risk of malignant spinal cord compression, compared to an 87% risk in patients with five or more risk factors. Back pain was nearly universal across the entire study population, and it failed to differentiate between those with and without metastatic cord compression. A cross-sectional study by Bayley et al examined factors that predicted subclinical spinal cord compression (ie, cord compression or thecal sac indentation without neurologic abnormalities) in patients with metastatic prostate cancer.[10] Using multivariate logistic regression analysis, the extent of disease score (determined from the number of lesions on a bone scan) and duration of hormonal therapy were the only factors predictive of subclinical cord compression (P = .02 and P = .04, respectively). Patients with extensive bone scan disease (> 20 metastases) had a 32% risk of malignant spinal cord compression prior to starting hormonal therapy, and they were at a 44% risk of developing cord compression after 24 months of hormonal therapy. Interstudy Comparisons
Bayley and colleagues' findings were consistent with those of Talcott et al in that back pain was not predictive of malignant spinal cord compression and in suggesting that patients with high-risk bone scans should be examined further in order to detect potential malignant spinal cord compression early. Ruckdeschel's algorithm, which uses "new back pain" as the heralding symptom, may indeed be predictive, but in light of these other two studies, the model should be evaluated in a prospective study. Talcott's risk estimates cannot be translated directly into others' practices without knowledge of baseline risk of cord compression. Using likelihood ratios calculated from Talcott et al's data,[9] the histology-specific incidence data from their populationbased study,[11] and Bayesian methodology, Loblaw and colleagues estimated the lifetime incidence of malignant spinal cord compression for different groups of asymptomatic patients (ie, no neurologic symptoms) according to their primary tumor site. Using these data, they were able to stratify the lifetime risk of cord compression in a patient over a 400-fold range. For neurologically intact patients with leukemia or ovarian, stomach, or pancreatic cancer and no Talcott risk factors, the risk of cord compression was negligible (0.05%). However, if a neurologically intact patient had prostate, female breast, or kidney cancer or myeloma and four Talcott risk factors, the estimated lifetime risk of malignant spinal cord compression was 19.3%. In the patient with newonset back pain without neurologic compromise, the pace-of-investigation algorithm proposed by Ruckdeschel could be modified based on a patient's baseline risk of having cord compression. Prevention of Malignant Spinal Cord Compression
If we can confirm these predictive models in prospective series, one might then envision different interventions designed to identify patients at high risk for compression and prevent symptoms from occurring. This could be done by administering systemic therapies (such as tetracycline(Drug information on tetracycline), bisphosphonates, better cytotoxic chemotherapies, or antiangiogenesis agents) earlier in the disease trajectory, or by identifying precursor lesions in the spine and treating them prophylactically before the patient becomes symptomatic. While research programs are actively investigating the systemic options, I believe the latter holds much promise-an intervention referred to as SMaRT (screening MRI and prophylactic radiotherapy). The Bayley study showed that thecal sac indentation can be identified in a large proportion of neurologically intact men with prostate cancer. All of these men with thecal sac indentation were treated with five fractions of radiotherapy. Although the study was not specifically designed to track the outcomes of these patients, all but one patient had no symptoms of spinal cord compression on follow- up. The one who did was successfully salvaged with surgery and remained ambulatory. Other authors show similarly low (average: 3%) rates of in-field failure after short courses of radiotherapy for cord compression.[ 7,10,12,13] However, while this study suggested that SMaRT was effective in preventing malignant spinal cord compresson, the magnetic resonance imaging (MRI) scans were only done at one point in time, patients were not followed in a standardized manner, and there was no comparator arm. There is therefore no information on the natural history of malignant cord compression (ie, what proportion of patients with subclinical cord compression would go on to develop symptoms of the complication) or on how frequently the MRIs should be repeated, and no data on important outcomes-namely functionality, quality of life (QOL), survival, cost, and health-care utilization. Investigators are now proposing a randomized study comparing SMaRT to best standard of care in men with prostate cancer at high risk of spinal cord compression. It is hypothesized that SMaRT may reduce the complications associated with malignant spinal cord compression and lead to lower health-care utilization and increased quality of life for these patients. Conclusions
Malignant spinal cord compression is a debilitating complication of cancer. New interventions are being evaluated, but fortunately, patients can experience much better outcomes if simple, proven strategies such as those discussed are followed.