|
|
| |||||||||||||||||||||||||||||||||||||||||
Background The standard follow-up care for children with medulloblastoma includes regular clinical evaluations and surveillance scanning of the central nervous system with computed tomography or magnetic resonance imaging. The evaluations and scanning assess the response of the tumor to treatment, detect any recurrence of disease, and monitor any complications of treatment. We compared the effectiveness of a periodic history taking and physical examination with that of surveillance scanning in detecting recurrent tumors.
Methods We reviewed the medical records, including 794 scanning reports or scans, of 86 children with posterior fossa medulloblastoma who were followed regularly between 1980 and 1991. Recurrent tumors were classified as symptomatic if neuroimaging studies had been prompted by clinical symptoms or signs and as radiographic if the tumor had been detected by imaging in an asymptomatic patient.
Results Twenty-three of the 86 children (27 percent) had a recurrence of tumor. Four recurrences (17 percent) were detected on scanning only, and 19 (83 percent) were associated with symptoms arising a median of four months after the previous scan. The median and range of survival after a recurrence of the tumor were 5 months and <1 to 38 months, respectively, for a symptomatic recurrence and 20 months and 10 to 32 months, respectively, for a radiographic recurrence (P = 0.03). No patient survived after a recurrence. The longer survival of patients with recurrent tumors detected by scanning most likely reflects the small number of patients and lead-time and length biases associated with screening.
Conclusions Among children with medulloblastoma, surveillance scanning is of little clinical value. Scanning detected a minority of recurrences, and no patient who had a recurrence survived.
Computed tomography (CT) and magnetic resonance imaging (MRI) are routinely used to diagnose brain tumors and to follow patients with such tumors during and after treatment. The primary purposes of this approach to surveillance are to assess the response of the tumor to treatment and to detect any recurrence of disease1. Surveillance scans complement a periodic history taking and physical examination. The practice of surveillance scanning is based on the assumptions that a tumor can recur without symptoms or long before symptoms appear and that detection of recurrent disease before the development of signs or symptoms may improve survival. However, surveillance scanning may require sedation or anesthesia and is costly.
At the Children's Hospital of Philadelphia and most other centers, children with medulloblastoma (posterior fossa primitive neuroectodermal tumor) are followed with the aggressive schedule of surveillance scanning recommended by Kun et al.1. We evaluated the ability of this schedule of surveillance scanning, as compared with a history and physical examination, to detect recurrent tumors.
Methods
The records and scans of 86 children with posterior fossa medulloblastoma diagnosed at Children's Hospital of Philadelphia between January 1, 1980, and December 31, 1991, formed the basis of this study. Patients with primitive neuroectodermal tumors outside the posterior fossa and those who had died within one month after diagnosis were excluded from the study2,3.
Between 1980 and 1985, the patients underwent contrast-enhanced CT scanning of the brain at the time of the diagnosis and CT scanning or CT myelography after surgery4,5,6. MRI was introduced in 1984. MRI scanning of the brain with gadolinium had largely supplanted CT scanning by 1988 and had replaced CT myelography by 19907,8. General anesthesia was required for CT myelography in the majority of children under the age of seven years and for MRI in about 10 percent of children under four years of age. Most children under the age of seven requested or required sedation for MRI studies.
After determination of disease stage, surgery, and further treatment (see below), surveillance CT or MRI scans of the brain were scheduled every three months for at least one year, then every six months for up to five years, and yearly thereafter. After 1988, surveillance MRI scans of the spine were scheduled every six months for patients whose initial postoperative scans showed residual tumor or whose cerebrospinal fluid obtained by lumbar puncture contained tumor cells and in children under five years of age who received less than the standard dose of radiation. An unscheduled scan was obtained if a patient had new or unexplained neurologic symptoms between scheduled evaluations.
A history was taken and a physical examination performed weekly during radiation treatment and at least three times every six weeks during chemotherapy; thereafter, the patients were examined at the time of scheduled scanning, unless additional examinations were indicated clinically. In comparing scanning with clinical evaluation, we counted only clinical evaluations performed as part of a planned schedule of surveillance or because of neurologic problems. The routine surveillance history taking and physical examinations were performed by neuro-oncologists or neurosurgeons in the brain-tumor clinic at Children's Hospital of Philadelphia.
Treatment varied according to the age of the patient, extent of surgical resection, and presence or absence of tumor dissemination9. Treatment consisted of surgery and radiation therapy or surgery and radiation therapy combined with adjuvant chemotherapy consisting of vincristine, prednisone and lomustine or vincristine, or cisplatin and lomustine, as previously described10,11,12,13. Recurrent tumors were treated with resection, radiation, multiple-agent chemotherapy, investigational agents, or only palliative measures14,15,16,17,18,19,20.
The medical records of each patient were reviewed to determine the extent of disease after surgery; the type, number, and results of surveillance scans; the results of history taking and physical examinations; the results of additional, unscheduled scanning performed because of clinical indications of recurrent disease; and compliance with the proposed follow-up plan. Only the scans obtained up to the time of a documented recurrence were included in the analysis of surveillance scanning; staging and diagnostic scans were excluded.
A recurrence of tumor was documented on the basis of a positive scan and a positive biopsy specimen, a positive scan and a clinical evaluation consistent with a recurrence, or a negative scan and a positive clinical evaluation plus a positive biopsy specimen. A scan was considered positive if it showed an enlargement of the primary tumor, a recurrence of the tumor in an area previously free of disease, or a metastasis. An increase in gadolinium enhancement alone was not considered evidence of a recurrence.
All comparisons were made with Fisher's exact test,21 except for the time from the diagnosis to the recurrence of disease and from recurrence to death, which was analyzed with the Wilcoxon rank-sum test22. Comparisons were two-tailed.
Results
There were 28 girls and 58 boys in the study, with a mean (±SD) age of 7 ±4 years (range, 0.5 to 21) at the time of the diagnosis. Fifty patients underwent a complete surgical resection and had no dissemination of the tumor, 17 underwent a partial resection with residual tumor in the posterior fossa but no dissemination, 10 had dissemination to the central nervous system without residual tumor, and 9 had both dissemination to the central nervous system and residual tumor at the primary site. Thirty patients received postoperative radiation therapy alone, and 56 received chemotherapy alone or chemotherapy and radiation therapy. The mean follow-up was 53 months (range, 4 to 158). Three patients did not comply with the follow-up recommendations.
Tumor Recurrence
Twenty-two patients had a recurrence of the tumor in the central nervous system; one patient had a recurrence in bone and bone marrow. The age and sex of the patients with and without recurrent tumors were comparable (Table 1). Those with a recurrence of disease were more likely to have residual primary tumor or central nervous system dissemination at the time of the diagnosis: 15 of the 23 (65 percent) with a recurrence had either or both of these features9,10,11,12,13. The sites of recurrent tumor are listed in Table 2. In six patients the tumor recurred locally in the posterior fossa. Five patients had a recurrence in brain parenchyma outside the primary site, and 11 had dissemination to the spinal cord or meninges. Among the patients with dissemination, two had a recurrence in the subarachnoid space in the spine without evidence of intracranial disease. The median time from diagnosis to recurrence was 13 months (range, 2 to 61).
|
|
A total of 794 surveillance scans were obtained: 485 CT scans of the brain, 234 MRI scans of the brain, and 75 scans of the spine (49 by CT myelography and 26 by MRI). Four (17 percent) of the recurrences were detected radiographically by surveillance scanning an average of 15 months after the diagnosis, and 19 (83 percent) were detected clinically because of symptoms an average of 17 months after the diagnosis (Table 3). The symptoms or signs that prompted unscheduled scanning included new or worsening ataxia, dysmetria, sixth-nerve palsy, diplopia, nystagmus, hemiparesis, lethargy, headaches, or early-morning vomiting; spinal or subarachnoid disease was signified by pain in the back, pain or weakness in the legs, hearing loss, or seizures. Percussion tenderness of the spine was the most common sign of spinal-cord disease.
|
CT scans confirmed the recurrence of disease in 10 of 18 symptomatic patients and diagnosed the recurrence of disease in 2 asymptomatic patients. A negative MRI surveillance scan preceded clinical signs of recurrence in seven symptomatic patients, and a positive MRI surveillance scan detected a recurrence in two asymptomatic patients.
Outcome
The median and mean (±SD) survival time after relapse was 5 months and 8 ±9 months (range, <1 to 38), respectively, in the patients who had symptomatic recurrences and 20 months and 21 ±10 months (range, 10 to 32) in those whose relapses were detected radiographically (P = 0.03) (Table 3). Treatment after a recurrence varied according to the previous therapy, the potential for resection, and the wishes of the patient and his or her parents. Two patients had excisional surgery, and one received only radiation therapy. Patients treated with chemotherapy received moderate-dose17 or high-dose16 cyclophosphamide, carboplatin,15 etoposide, etoposide plus ifosfamide,20 vincristine, lomustine with either cisplatin or procarbazine,14,19 or a regimen of eight drugs in one day18. Five patients did not receive specific antitumor therapy after a recurrence of the tumor had been documented. No patient with recurrent disease survived.
Discussion
Among 86 children with medulloblastomas, surveillance scanning detected 4 asymptomatic recurrent tumors (17 percent) among a total of 23 recurrences. Symptoms or signs heralded or coincided with a recurrence in 19 (83 percent) of the patients with a recurrence. No patient with recurrent tumor was cured. Indeed, a cure of recurrent medulloblastoma is rare14,15,16,17,18,19,20,23,24. The median survival of the 4 asymptomatic patients in whom a recurrence was detected by scanning was 15 months longer than that of the 19 patients who had symptomatic recurrences. Among 21 patients with recurrent medulloblastoma in another study, the 13 in whom a recurrence was detected by surveillance scanning had a longer mean survival than the 8 who had symptomatic recurrences25. The gain in survival probably reflects a lead-time bias because of early detection. It may also reflect a length bias; that is, as compared with patients whose disease is detected clinically, those with disease detected by scanning are likely to have tumors that grow more slowly or in clinically less important areas26,27.
The study period encompassed an era when MRI was replacing CT as a means of surveillance. We cannot rule out the possibility that MRI would have detected more tumors sooner. MRI might also have detected the marrow or spinal cord disease in the three patients with false negative CT scans of the spine4,7,8. It may therefore be necessary to validate these results in a prospective study using gadolinium-enhanced MRI.
The results of this study are similar to those of surveillance bone marrow examinations in children with acute lymphoblastic leukemia. Among 1466 such children, 6890 surveillance examinations detected 43 subclinical relapses (19 percent) among a total of 221 relapses28. Surveillance detected a relapse about one month before symptoms occurred but had no effect on survival28. The use of serum carcinoembryonic antigen measurements to detect recurrent tumors in adults with resected colon cancer has had similar results: the test detected recurrent disease in 59 percent of patients, of whom only 3 percent were alive and disease-free one year later, as compared with 2 percent of those who did not undergo testing29. Surveillance practices in patients with medulloblastoma, acute lymphoblastic leukemia, and colon cancer have been based on the assumption that the detection of minimal recurrent disease provides a greater opportunity for treatment. However, in these three diseases, eradication of resistant tumor cells has proved nearly impossible, no matter how small their number.
We conclude that surveillance scanning in children with medulloblastoma has limited clinical value. Increasing the frequency of scanning is neither practical nor likely to change the outcome. The disadvantages of more frequent scanning include the cost, the morbidity associated with sedation or general anesthesia in young children, the competition for scanning services, the fact that patients are unlikely to be cured of recurrent disease, and the probability that any increase in survival is an artifact of lead-time and length biases.
Our conclusion about the value of scanning may not be valid if the technique becomes more sensitive. However, greater sensitivity would also bring less specificity. The conclusion may also change if a cure is developed for recurrent medulloblastoma. At present, we recommend regular clinical evaluation and regular scanning to determine the maximal tumor response and to detect a change in status after therapy. Thereafter, scanning should be based on clinical indications.
Supported in part by grants from the National Institute of Neurologic Diseases and Stroke (NS31102, to Drs. Phillips and Lange) and the National Cancer Institute (KO4 CA01480, to Dr. Silber).
We are indebted to the Greater Delaware Valley Tumor Registry for assisting in data collection and to Ms. Diane Howard for assistance in the preparation of the manuscript.
Source Information
From the Departments of Neurology (C.F.T., S.R., P.C.P.), Neurosurgery (L.N.S.), Radiology (L.T.B., R.A.Z.), and Pediatrics (B.J.L.), Children's Hospital of Philadelphia; the Leonard Davis Institute of Health Economics and Department of Pediatrics (J.H.S.); and the Department of Radiation Therapy, Hospital of the University of Pennsylvania (J.W.G.) -- all at the University of Pennsylvania School of Medicine, Philadelphia.
Address reprint requests to Dr. Lange at the Division of Oncology, Department of Pediatrics, Children's Hospital of Philadelphia, University of Pennsylvania School of Medicine, Philadelphia, PA 19104.
References
| |||||||||||||||||||||||||||||||||||||||||
Related Letters:
Surveillance Scanning of Children with Medulloblastoma
Lindsley K. L., Lange B. J., Phillips P.
Extract |
Full Text
N Engl J Med 1994;
331:483, Aug 18, 1994.
Correspondence
More on Surveillance of Children with Medulloblastoma
Friedman H. S., Kun L. E.
Extract |
Full Text
N Engl J Med 1995;
332:191, Jan 19, 1995.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
