|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2001 American Society for Clinical Oncology Surveillance Neuroimaging to Detect Relapse in Childhood Brain Tumors: A Pediatric Oncology Group StudyByFrom the Departments of Neurology, Pediatrics, and Health Research and Policy, Stanford University, Palo Alto; Department of Pediatrics, University of California, Davis, CA; Department of Health Policy and Epidemiology, University of Florida and Pediatric Oncology Group Statistical Office, Gainesville, FL; and Department of Radiation Oncology, St Jude Children’s Research Hospital, Memphis, TN. Address reprint requests to Paul G. Fisher, MD, Rm A343, Department of Neurology, Stanford University School of Medicine, 300 Pasteur Dr, Palo Alto, CA 94305-5235; email: pfisher{at}stanford.edu
PURPOSE: To investigate the prognostic significance of surveillance neuroimaging for detection of relapse among children with malignant brain tumors. PATIENTS AND METHODS: A historical cohort study examined all children who experienced relapse from 1985 to 1999 on one of 10 Pediatric Oncology Group trials for malignant glioma, medulloblastoma, or ependymoma.
RESULTS: For all 291 patients (median age at diagnosis, 8.2 years), median time to first relapse was 8.8 months (range, 0.6 to 115.6 months). Ninety-nine relapses were radiographic, and 192, clinical; median time to relapse was 15.7 versus 6.6 months, respectively (P = .0001). When stratified by pathology, radiographic and clinical groups showed differences in median time to relapse for malignant glioma (7.8 v 4.3 months, respectively; P = .041) and medulloblastoma (23.6 v 8.9 months, respectively; P = .0006) but not ependymoma (19.5 v 13.3 months, respectively; P = .19). When stratified by early (< 8.8 months) or late ( CONCLUSION: Surveillance neuroimaging detects a proportion of asymptomatic relapses, particularly late relapses, and may provide lead time for other therapies on investigational trials. During the first year after diagnosis, radiographic detection of asymptomatic relapse was infrequent. A prospective study is needed to formulate a rational surveillance schedule based on the biologic behavior of these tumors.
SURVEILLANCE neuroimaging by magnetic resonance imaging (MRI) and computed tomography (CT) has become standard practice to detect brain tumor relapse in children, although the timing of imaging is arbitrary. Despite the differences in natural history for each tumor pathology, there is little variation in the recommended schedule of scans.1,2 In addition, for medulloblastoma, controversy has surfaced about whether surveillance scanning confers any prognostic benefit.3,4 Ideally, a timetable for neuroimaging should identify recurrent or progressive tumors at an early stage, when additional therapy may be effective. A better understanding of the natural history and failure patterns of the different tumor types should lead to a more efficient surveillance schedule and strategies designed to detect relapse, implement salvage therapy, and improve survival. This study sought to define further the failure patterns of malignant brain tumors in children and to examine the prognostic importance of current surveillance neuroimaging schedules to detect relapse.
Patient Eligibility and Record Review To assemble a complete, consecutive cohort of children with malignant brain tumors who experienced first relapse, the database of the Pediatric Oncology Group (POG) Statistical Office in Gainesville, Florida, was reviewed. Patients considered for study were children aged 3 through 21 years at initial diagnosis of a malignant brain tumor enrolled onto one of 10 POG therapeutic protocols between 1985 and 1999. Eligible pathologies included medulloblastoma (POG 8631, 8695, 9031, and 9331), ependymoma (POG 8532, 9132, and 9432), and high-grade glioma (POG 8832, 9135, and 9431), such as glioblastoma multiforme, anaplastic astrocytoma, anaplastic oligodendroglioma, and gliosarcoma, or, rarely, poorly differentiated embryonal tumors (14 patients). This latter group of tumors is hereinafter referred to as malignant gliomas. Infants less than 3 years of age at diagnosis were excluded because of the known heterogeneous pathologies and short time to relapse.5 In addition, patients with brainstem glioma were excluded because of the exceedingly poor survival rates and difficulty defining the relapse event. Of 722 patients enrolled onto these protocols, 328 patients experienced a documented first relapse. We were able to ascertain the mode of relapse assessment for the 291 children who comprised the study cohort. Protocol flow sheets and other medical records were reviewed to abstract date of birth, sex, race/ethnicity, date of diagnosis, primary tumor site, extent of surgical resection, date of relapse, mode of detection, date of neuroimaging at relapse, survival status, and cause of death.
Assessment of Relapse Relapse was defined as the reappearance or progression of tumor at the original site or the appearance of tumor at sites previously free of disease.6 All relapses prompted a change in treatment plan, ie, removal from the clinical trial. Death because of tumor was considered a relapse when it was evident that the tumor had not responded to treatment. Relapses were classified by mode of detection: radiographic (ie, asymptomatic) or clinical (ie, symptomatic). Clinical relapse was defined by symptoms (eg, change in ophthalmologic, endocrinologic, or neurologic status) that indicated clear recurrence or progression, and confirmed by changes in MRI, CT, or CSF cytology. Radiographic relapse was defined as an asymptomatic increase of more than 25% in tumor volume, a recurrence of tumor in a region previously free of disease, or a new metastasis detected by routine surveillance MRI or CT. A change in the pattern of contrast enhancement alone was not considered indicative of relapse.
Statistical Analysis
Characteristics of the 291 patients whose mode of relapse detection could be determined are listed in Table 1. Malignant glioma patients accounted for 50.2% of the cohort. The predominant race/ethnicity of children was non-Hispanic white (70.0%). Of the relapses, 192 (66.0%) were detected clinically, and 99 (34.0%) were detected radiographically.
Time to relapse was significantly shorter for the clinically detected group (Kruskal-Wallis P = .0001) (Table 2). When time to relapse was dichotomized as early (< 8.8 months) and late (  8.8 months), a greater proportion of early relapses were detected clinically. This remained significant, even after adjustment for pathology (Mantel-Haenszel P = .001) Age at initial diagnosis, sex, and race were not significantly associated with the likelihood of experiencing a radiographically versus clinically detected relapse. The radiographic group was more likely to have greater initial extent of surgery as compared with the clinical group (Kruskal-Wallis P = .04). However, this relationship did not retain significance after adjustment for pathology (Mantel-Haenszel P = .19).
Patient characteristics by pathology are listed in Table 1. For all tumor pathologies, there were no statistically significant differences between radiographic and clinical detection groups with respect to age at initial diagnosis, sex, race, primary tumor site, or extent of surgical resection. When stratified by pathology, time to relapse was significantly different whether determined radiographically or clinically for malignant glioma and medulloblastoma, but not ependymoma (Table 3). For medulloblastoma, no difference between relapse groups was observed with respect to initial Chang metastatic stage.8 Insufficient staging data were available for other tumor types.
The 1-year OS rate ± SE from relapse for all cases was 37% ± 3%. OS from relapse was significantly different among tumor types (log-rank P = .002), with malignant glioma patients having the poorest outcome. One-year OS rates from relapse for malignant glioma, ependymoma, and medulloblastoma were 25% ± 4%, 51% ± 8%, and 36% ± 5%, respectively. OS from relapse among children detected clinically was significantly shorter (log-rank P = .002) (Fig 1). Median survival from clinical relapse was 5.5 versus 10.8 months for radiographic relapse. The 1-year OS rate from relapse for the clinical group was 33% ± 3%, compared with 46% ± 5% for the radiographic group. Among children who experienced early relapse, OS was not significantly different whether relapse was detected clinically or radiographically. However, among patients who experienced late relapses, 1-year OS rates were different for the two detection modes (clinical, 27% ± 5%; radiographic, 49% ± 6%; log-rank P = .003).
When OS from relapse by mode of detection was analyzed according to tumor types, the trend for improved OS among radiographic relapses continued but did not retain significance. Among gliomas, 1-year OS for the clinical group was 25% ± 4%, compared with 30% ± 7% for the radiographic group (log-rank P = .11) (Fig 2A). Among ependymoma, 1-year OS from relapse for the clinical group was 55% ± 11% versus 58% ± 10% for the radiographic group (log-rank P = .10) (Fig 2B). In medulloblastoma, 1-year OS from relapse for the clinical group was 38% ± 6%, in contrast to 56% ± 8% for the radiographic group (log-rank P = .12) (Fig 2C).
A greater extent of surgical resection at initial diagnosis was positively associated with improved survival even after relapse (1-year OS: gross total resection, 42% ± 8%; subtotal, 36% ± 5%; partial, 34% ± 7%; biopsy, 27% ± 6%; log-rank P = .05). Initial extent of resection was also associated with pathology (  2 P = .001). There were no statistically significant associations between survival after relapse and sex, race, or primary tumor site.
Although the majority of relapses on POG trials were determined clinically, a large proportion were detected asymptomatically, particularly late relapses. A previous single-institution study found radiographic recurrence to be rare, although CT was the principal mode of surveillance.3 MRI, as performed on the POG trials, may afford more sensitivity to detect relapse. Regardless, routine neuroimaging did not detect the majority of relapses, contrary to another limited single-institution review.4 Unlike the POG experience, patients in the latter study underwent MRI every 3 months for the first 2 years after diagnosis, then every 6 months for the next 3 years. Our observational study suggests that surveillance neuroimaging confers a small but prognostic advantage in survival from the time of relapse. This finding could be interpreted as intuitive, as surveillance screening produces a lead-time bias in detection of asymptomatic relapse. However, in our experience, time to relapse was, on average, longer when detected radiographically rather than clinically. Detection of asymptomatic relapse may be valuable for the administration of salvage therapy and the evaluation of highly experimental approaches in phase I and II trials. Indeed, families and children may be more willing to participate in innovative therapies they would later defer once the patient becomes symptomatic. Neuroimaging of children with brain tumors has many facets not considered in this study. First, MRI is often necessary after 2 or 3 months of treatment to measure response, particularly on experimental chemotherapy protocols. Secondly, imaging may be indicated at certain intervals to reassure patients and their families or to monitor for complications of therapy. Finally, surveillance is not without costs. MRI in children can be expensive, requires sedation or anesthesia, and may cause unnecessary psychologic stress for the patient or family. There are several limitations to this investigation. First, we used data collected primarily for other purposes. The level of detail about the means by which relapse was detected may not have been complete. We could not measure compliance with the recommended schedule for surveillance scans. In fact, the mode of relapse detection could not be discerned for 35 patients. These patients were excluded from analysis. Secondly, because this was not a randomized study, there may have been residual confounding by unknown covariates. However, we stratified and adjusted the analysis by known confounders. Third, CT imaging, rather than MRI, was positively associated with clinical relapse, not radiographic relapse. All children were nevertheless receiving routine surveillance MRI, but some likely underwent CT at clinical relapse because this method is quicker and more widely available for a neurologically ill patient. Fourth, one could argue that late relapses are less biologically aggressive, and MRI detection of such is logically associated with increased survival. However, when we restricted analysis just to late relapses, OS from relapse was still significantly improved, comparing radiographic with clinical relapse. Finally, selection bias might be present, as children with the extremes in prognosis are sometimes not enrolled onto clinical trials. However, these cooperative group data account for nearly a quarter of all children in North America with these tumor malignancies diagnosed over a 10-year period. Our findings represent the first large-scale study to show that surveillance imaging for relapse clearly seems to have clinical utility in brain tumor patients. Thus, reconsideration of neuroimaging strategies using surveillance MRI is reasonable. The schedule used by POG3 differs from the routine recommended by the Children’s Cancer Group.4 For the pathologies examined in this study, Children’s Cancer Group recommended MRI at 3, 6, 9, 12, 16, 20, 24, 30, 36, 48, and 60 months from diagnosis. In light of the number of relapses that occur over a year from diagnosis, more frequent scanning during the second and third years may be efficacious. We did not find many relapses in the first 9 months from diagnosis by surveillance MRI, likely because these early symptomatic relapses represent more aggressive disease. Whether the frequency of MRI should be decreased or increased in the first year is debatable. Furthermore, the differing times to relapse among the pathologies studied suggest that neuroimaging protocols may need to vary for the different tumor types. Prospective study of neuroimaging surveillance is warranted for future cooperative group trials. Such a study would provide the evidence needed to guide the design of appropriate surveillance strategies.
Supported by the Mihos Foundation, Boston, MA, and the Pediatric Oncology Group Foundation, Chicago, IL (A.Y.M.), and in part by grant no. K12NS01692 from the National Institutes of Health/National Institute of Neurological Disorders and Stroke, Bethesda, MD (P.G.F.).
Presented in part at the Thirty-Sixth Annual Meeting of the American Society of Clinical Oncology, New Orleans, LA, May 20-23, 2000.
1. Kun LE, D’Souza B, Tefft M: The value of surveillance testing in childhood brain tumors. Cancer 56: 1818-1823, 1985[Medline] 2. Kramer ED, Vezina LG, Packer RJ, et al: Staging and surveillance of children with central nervous system neoplasms: Recommendations of the Neurology and Tumor Imaging Committees of the Children’s Cancer Group. Pediatr Neurosurg 20: 254-262, 1994[Medline]
3.
Torres CF, Rebsamen S, Silber JH, et al: Surveillance scanning of children with medulloblastoma. N Engl J Med 330: 892-895, 1994
4.
Shaw DW, Geyer JR, Berger MS, et al: Asymptomatic recurrence detection with surveillance scanning in children with medulloblastoma. J Clin Oncol 15: 1811-1813, 1997 5. Fisher PG, Needle MN, Cnaan A, et al: Salvage therapy after postoperative chemotherapy for primary brain tumors in infants and very young children. Cancer 83: 566-574, 1998[Medline] 6. Zeltzer PM, Friedman HS, Norris DG, et al: Criteria and definitions for response and relapse in children with brain tumors. Cancer 56: 1824-1826, 1985[Medline] 7. Peto R, Pike MC, Armitage P, et al: Design and analysis of randomized clinical trials requiring prolonged observation of each patient: Analysis and examples. Br J Cancer 35: 1-39, 1977[Medline] 8. Chang CH, Housepian EM, Herbert C: An operative staging system and megavoltage radiotherapeutic technic for cerebellar medulloblastomas. Radiology 93: 1351-1359, 1969[Medline] Submitted August 11, 2000; accepted May 15, 2001.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2001 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
