Long-Term Survival of Children with Brain Tumors

Long-Term Survival of Children with Brain Tumors

The diagnosis and treatment of children with brain tumors has changed radically over the last 50 years. Cross-sectional imaging, CT and MRI, has displaced angiography and pneumoencephalography. These newer imaging modalities have facilitated early diagnosis, preoperative planning, and surgical approach, resulting in an increased likelihood of achieving complete surgical extirpation. The operating microscope has improved the experienced surgeon's ability to discriminate between tumor and normal brain, making radical resection more frequent. Chemotherapy has been introduced into the arsenal of the neuro-oncologist, albeit with only modest success. The one nearly constant treatment modality has been external-beam irradiation.

Dr. Jenkin's description of the Toronto experience offers a unique opportunity to review a consistent single-institutional experience over 38 years with a wealth of follow-up data. These data are particularly enlightening, in terms of both providing insights into the long-term outcome of these patients and pointing out the limitations of local therapy. One of the greatest limitations of this data set is the attempt to uncover prognostic factors in a heterogeneous population. This group has varied histologies and varied treatments, both between and most likely within tumor types. Therefore, it is possible that an important treatment advance for one histology may be lost in the larger data set.

The raw data on survival demonstrate vividly the prognostic value of histology. These data not only demonstrate that optic pathway glioma has the best prognosis and high-grade glioma the worst, but also show that, for the lower-grade histologies (optic glioma and low-grade astrocytoma), late mortality from disease is a reality. In contrast, patients with medulloblastoma and high-grade glioma who are going to relapse are most likely to do so early, within the first 5 years.

Haunting Data on Second Malignancies

The author's Figure 6 provides haunting data on the risk of second malignancies. By 30 years after treatment, the rate of second malignant neoplasms was 19%, without any evidence that this trend will level off. One could envision data suggesting that any patient who lives long enough will eventually develop a second tumor. Calling all second tumors malignant may be misleading. Nine patients had a second glioma, but no data are given on the grade of these tumors. Seven patients had a meningioma; most of these tumors are likely to be benign histologically and can be remedied with surgery.

The 58% rate of 5-year survival following diagnosis of second tumors suggests that many patients had tumors that were curable with surgery alone, as reirradiation usually is not an option. Most notable is the incidence of acute leukemia (histology unspecified) after treatment for brain tumors. Although the risk of secondary brain tumors following cranial irradiation for central nervous system prophylaxis is well known, reports of secondary leukemia, particularly lymphoblastic leukemia, are scarce. Not surprisingly, all patients with secondary leukemia in this study died, as do children with secondary leukemia from other therapies.

Two Important Points About Radical Excision

One of the most controversial aspects in the treatment of childhood brain tumors is the value of radical excision. The author's Figure 3 demonstrates two important points. First, radical excision is important for tumor control and survival at 5 years, the end point for most studies. However, even more impressive is the benefit for the long-term survivor. Whereas the mortality curve for the radically resected patient is flat at year 10, the subtotally resected patient is still at risk for late mortality. For patients who died beyond 10 years from diagnosis, 20 died from the primary tumor and 8 from a second malignant neoplasm.

The lack of late mortality in the completely excised group suggests a lower risk of secondary tumors, which account for 28% of late deaths. This is hard to explain, unless these are patients who had reduced-dose irradiation. Of course, any analysis of the extent of surgical resection is incomplete without a discussion of surgically induced neurologic morbidity. One would hope that, with improved imaging and operative technique, there would be little change in morbidity despite a constantly rising rate of total resection, from 17% to 31%.

Cognitive Outcome Not Addressed

Another issue that is not addressed in this study is the cognitive outcome of the patients. This population presents a unique opportunity to measure long-term morbidity. What was their highest level of education? Are these patients working? Married? The data presented here clearly demonstrate that survival is unaffected by age, with patients less than 4 years of age doing no worse than older patients. These findings do not take into account data from other investigators, which clearly demonstrate the negative impact of whole-brain irradiation on cognition, virtually preventing its use in the youngest patients. To fully assess the benefit of any therapy, one must look at both its morbidity and mortality in comparison to alternative measures.

Dr. Jenkin notes the lack of statistically significant improvement in survival over the course of the study, and suggests that since some low-grade tumors were no longer irradiated in the later years of study, the more recent patient population may have had more aggressive disease. I would prefer to look at the same hypothesis from another vantage point. Some of the patients with completely resected cerebellar astrocytoma in the earlier era were surgically cured prior to referral to radiation oncology. Attributing long-term survival in these patients to irradiation may inflate our perception of the efficacy of this modality.

Currently, we (in Philadelphia) are irradiating patients with completely excised ependymoma. The data suggesting that extent of resection is the most important prognostic sign after presence of dissemination [1] lead one to question whether these patients need irradiation at all. This is one example in which an attempt to analyze data that include multiple histologies can be misleading.

Some Incremental Improvement

Dr. Jenkin makes some strong summary statements in the abstract to his paper: (1) Long-term survival has not improved substantially over the last 38 years; and (2) chemotherapy is largely experimental. The data he presents do demonstrate a lack of improvement in survival over the duration of the study. But lumping all histologies together covers up some incremental improvement. Survival in higher-risk medulloblastoma has improved with the addition of cisplatin (Platinol)-based chemotherapy [2]. Although survival has been unaffected, use of chemotherapy to postpone irradiation in the youngest patients with malignant brain tumors has substantially reduced morbidity [3]. The fact that some infants can achieve long-term progression-free survival without radiation demonstrates the potential efficacy of chemotherapy.

Where to go from here? In the face of the data presented by Dr. Jenkin, there appears to be little hope in improving the survival for children with brain tumors without developing increasingly effective systemic therapies. For all children with brain tumors, except those with low-grade astrocytoma, which can be completely excised, half will succumb to tumor, and the survivors face at least a 19% risk of a second tumor.


1. Healey EA, Barnes PD, Kupsky WJ, et al: The prognostic significance of postoperative residual tumor in ependymoma. Neurosurgery 28:666, 1991.

2. Packer RJ, Sutton LN, Elterman R, et al: Outcome for children with medulloblastoma treated with radiation and cisplatin, CCNU, vincristine chemotherapy. J Neurosurg 81:690, 1994.

3. Duffner PK, Horowitz ME, Krischner JP, et al: Postoperative chemotherapy and delayed radiation in children less than three years of age with malignant brain tumors. N Engl J Med 328:1725, 1993.

Loading comments...
Please Wait 20 seconds or click here to close