Low-grade astrocytomas in children: evolving management strategies

Abstract

Developments in imaging and in neurosurgical techniques over the past decade have substantially altered the management of children with low-grade astrocytoma. Indications for surgery have become more clearly defined, and a larger proportion of children undergo complete or subtotal resection than in the past. Fewer receive adjuvant therapy, even though the options in this regard are more numerous now and theoretically likely to result in less morbidity than conventional external beam radiotherapy.

This review will address in particular the correlations between location, imaging appearance, and behavior that need to be more widely appreciated, and present recommendations regarding the management of these tumors.

Introduction

Low-grade astrocytomas (LGA) account for one-third to one-half of all central nervous system (CNS) tumors in the pediatric age group. They are a heterogeneous group of tumors whose principal common characteristic is that of an indolent clinical course, with overall survival rates at 10 and 15 years as high as 80 to 100%. Cerebellar astrocytomas are the most prevalent (15–25% of all CNS tumors), followed by hemispheric astrocytomas (10–15% of all CNS tumors), tumors of the deep midline structures including the corpus callosum, lateral and third ventricles, the hypothalamus, and thalamus (10–15% of all CNS tumors), and tumors of the visual pathway (approximately 5% of all CNS tumors). Tumors arising in the brain stem account for a further 10 to 15% of all CNS tumors; 15–20% of these, characterized by a typical clinical presentation and radiological appearance, are low-grade astrocytomas. Finally, approximately 70% of spinal cord tumors (3–6% of all CNS tumors) are low-grade astrocytomas. In practice, either because of a specific pattern of behavior or because of constraints on management due to location, it is necessary to consider several distinct entities at most of these sites.

Over the past decade there have been tremendous advances in a number of spheres that directly impact on the care of children with LGA. These include the advent of improved imaging, especially magnetic resonance (MR), improved neurosurgical techniques, and perioperative care, as well as improved radiotherapy planning and treatment techniques. Issues concerning the management of these tumors that will be addressed in this review include:

• the need for greater appreciation of the different patterns of behavior of the various tumor types and the indication for and the timing of any therapeutic intervention,

• the results of treatment with surgery alone in the context of increasingly aggressive surgical resections,

• the role of adjuvant therapy (radiotherapy or chemotherapy) after surgical resection and the developments, particularly in radiotherapy that may influence the choice of modality.

Most LGA of children are of one of two distinct types, namely pilocytic astrocytoma, and low-grade astrocytoma of so-called diffuse or fibrillary type. According to the new WHO classification (Table 1), pilocytic astrocytomas are designated as Grade I, regardless of any specific histologic characteristic that in other classification systems (Kernohan, Ringertz, Ste. Anne/Mayo) would lead to their designation as higher grade lesions. All other low-grade astrocytomas are designated as Grade II in the WHO classification. That these two types of LGA should be regarded as distinct entities is supported by molecular genetic studies that show genetic alterations including p53 mutations and loss of heterozygosity on chromosome 17p in the latter (in common with higher grade astrocytomas) but not in pilocytic astrocytomas (1).

Pilocytic astrocytomas are the most common type in the pediatric age group. Occurring predominantly in young children (median 4 years), they account for almost all of the LGA at certain sites, for example, optic nerve tumors and dorsal exophytic brain stem tumors, and the majority (85%) of LGA in the cerebellum. They account for a smaller proportion of LGA arising in the deep midline structures and in the cerebral hemispheres. Grossly, pilocytic astrocytomas appear well-circumscribed and frequently, particularly in the cerebellum, partly cystic. Microscopically, they are composed in varying proportions of aligned bipolar astrocytes, often associated with Rosenthal fibers, and loose-knit microcystic areas that contain stellate astrocytes in a mucoid matrix. Vascular proliferation is common and the blood vessels may be markedly hyalinized (2). It appears that some degree of infiltration into surrounding brain tissue at the tumor margin may be quite common (3), in spite of the fact that on imaging these tumors are almost always well demarcated from surrounding brain tissue, with little or no edema and no mass effect, and at surgery separate easily from normal brain.

Low-grade astrocytomas of the so-called diffuse or fibrillary type occur at a somewhat older age (median 10 years). Most intrinsic pontine tumors and a large proportion of astrocytomas arising in the cerebral hemispheres are fibrillary astrocytomas. They are more densely cellular and more widely infiltrative, grossly and microscopically, and tumor cells can often be identified in surrounding normal brain 2, 4. In contrast to pilocytic astrocytomas, fibrillary astrocytomas may undergo malignant degeneration, although this is an uncommon event in the pediatric age group.

Patients with LGA usually present with a long history of nonspecific and nonlocalizing symptoms. Symptoms and signs of raised intracranial pressure are common in LGA at all sites but especially in midline and cerebellar tumors, which are more likely to obstruct CSF circulation. Posterior fossa tumors may present with neck stiffness and head tilt as a manifestation of intracranial hypertension causing tonsillar herniation, altitudinal diplopia, or spinal accessory nerve herniation. Seizures are present in as many as three-quarters of patients with hemispheric tumors.

Other symptoms, relatively less frequent and usually of more recent onset, relate to the location of the tumor. These may include, for example, focal motor deficits with hemispheric tumors, the diencephalic syndrome with diencephalic tumors, neuroendocrine deficits with hypothalamic tumors, and loss of vision with tumors compressing or involving the visual pathway.

On nonenhanced computerized tomography (CT) scans, most LGA are hypodense. Typically, pilocytic astrocytomas enhance brightly and uniformly with contrast material (Fig. 1) while fibrillary astrocytomas usually show little enhancement with contrast material. MR delineates more precisely the extent of disease and, particularly for fibrillary astrocytomas, better distinguishes the tumor from uninvolved surrounding normal tissues (4) (Fig. 2). LGA are usually hypointense on T1- and hyperintense on T2-weighted images. Gadolinium enhancement patterns parallel enhancement patterns seen with CT.

Specific imaging characteristics are increasingly appreciated as a guide to therapy. In general, a well-defined border on T2-weighted MR, bright gadolinium enhancement, and the presence of an associated cyst are all predictive of pilocytic histology and of a well-defined interface between tumor and surrounding tissues, and favor a more aggressive surgical approach, even in eloquent areas such as the thalamus or brain stem (Fig. 3).

Published data with regard to the management of LGA are extremely difficult to interpret. Not only do the majority of reported series include both adult and pediatric cases, and in many instances all sites, but most are also selected in that they are based on personal experiences or referral patterns. Many include patients treated over several decades, during which time there have been tremendous developments in diagnostic and therapeutic techniques. Nonetheless, some statements with regard to the management of children with LGA may be made as follows:

• Some children with LGA may not require any treatment. These include, for example, (i) patients with neurofibromatosis with visual pathway or other tumors that are asymptomatic and nonprogressive, and (ii) patients with tectal tumors who present with only hydrocephalus who may do well after CSF diversionary procedures alone. As many as 20% of children with neurofibromatosis type 1 have optic pathway tumors (5). For the most part, at least since the advent of CT and MR imaging, these are found incidentally, and patients may remain asymptomatic with nonprogressive lesions over long periods of time. The actual frequency of progression is difficult to establish. In one recently reported series only 9% of patients showed evidence of tumor growth during the study period, which was, however, relatively brief (median 2.4 years for asymptomatic patients) (5). In a recently completed Pediatric Oncology Group (POG) study whose principal objective was to study the natural history of visual pathway tumors, 23 of the 59 patients with neurofibromatosis entered on study remained progression free without treatment after a median follow-up interval of 38 months (M. Cohen, personal communication, November 1997). Progression-free survival without treatment may be even better for patients with tectal tumors who present with hydrocephalus without localizing brain stem signs. In two recently reported studies all such patients managed without any tumor-specific treatment remained free from progression after median intervals of 3$\text{1}\text{2}$ and 5 years 6, 7. In a third series of patients followed for longer periods of time, 4 of 16 patients progressed at a median of 7.8 years from shunt placement and 11.5 years from onset of symptoms. All four were salvaged with radiotherapy at time of progression (8). Clearly, these are small series of highly selected patients. Nonetheless, the extremely favorable outcome points to the need for careful evaluation and individualization of management, depending on the specific clinical situation and tumor type.

• Surgery is the mainstay of treatment for LGA. Gross total resections are more likely to be accomplished in patients with smaller tumors and those arising in noneloquent sites as well as in patients with the generally well-circumscribed pilocytic tumors, factors that, as noted previously, are to a large extent interrelated. Complete resection rates are estimated as approximately 80% for cerebral and cerebellar tumors and 40% for diencephalic tumors. Modern surgical techniques that include, for example, computer-assisted resections aided by electrophysiologic mapping of eloquent areas may permit greater degrees of resection in larger proportions of patients 9, 10, including many who until relatively recently would have been considered to have inoperable lesions. Complete surgical resection leads to long-term disease-free and overall survival in 80 to 100% of children 11, 12, 13, 14, 15, 16, 17, 18. In most series, results are better (i.e., close to 100%) for patients with pilocytic astrocytomas than for those with fibrillary astrocytomas 11, 16, 18, 19, 20, 21, although some have disputed this, showing that results may be equally satisfactory for patients with fibrillary astrocytomas 8, 22. In either type, postoperative adjuvant therapy is clearly not indicated. Children who undergo less than complete resection fare less well. For such patients progression-free survival rates after surgery alone range from 33 to 67%. However, a significant proportion of patients can be salvaged with a second surgical resection and/or radiotherapy, and overall survival at 5 to 10 years ranges from 50 to 94% for hemispheric and cerebellar LGA 12, 15, 16, 17, 23. The impact of degree of resection (i.e., subtotal vs. partial vs. biopsy) is difficult to assess because of confounding variables most especially with regard to tumor size, location, and histologic type. However, in general, patients who have undergone subtotal (>90%) resection are thought to have a prognosis after surgery alone not greatly different from those who have undergone complete resection, and the usual recommendation following surgery will be close follow-up without further intervention.

• The role of postoperative radiotherapy following incomplete tumor resection remains unclear. In spite of the obvious biases inherent in retrospective studies in which less favorable patients would have been treated with radiotherapy, the use of radiotherapy following less than complete resection is reported in several series to result in better disease-free survival 12, 15, 16, 17. Overall survival rates for children with hemispheric and cerebellar LGA treated with less than complete resection followed by radiotherapy range from 67 to 100% at 5 years 12, 16, 24, 25 and 63 to 89% at 10 years 17, 24, 25. However, given the valid concerns with regard to the use of radiotherapy in young children, the role of postoperative radiotherapy needs to be reassessed in the context of modern surgical techniques and improved imaging for follow-up. Such was the principal objective of a recently completed joint Children’s Cancer Group (CCG)-POG study. More than 700 patients were accrued to this study, the results of which are not yet available. At the present time, the usual recommendation for a patient who has undergone less than complete resection and who is neurologically stable is close follow-up (Fig. 4). A second surgical procedure would be considered at time of progression, and radiotherapy reserved for patients with progressive, inoperable disease.

• The optimal management of patients with deep midline and other tumors previously considered surgically inaccessible remains a matter for debate. These patients have traditionally been treated with radiotherapy, often without histological confirmation of diagnosis. Using such an approach, progression-free survival rates as high as 75% at 10 years have been reported (26). Overall survival rates range from 33 to 75% at 10 years 25, 26, 27, 28, 29, 30, and in one population-based study, 64% at 15 years (30). Using newer neurosurgical techniques several groups have shown that it is feasible to surgicaly resect at least some of these lesions 9, 31. However, whether the results are better than those achieved utilizing a more conservative approach remains to be established (32). Moreover, newer radiotherapy techniques such as conformal radiotherapy, brachytherapy, or stereotactic irradiation may prove to be excellent alternatives for the same group of patients, that is, those with well-circumscribed and relatively small-volume tumors (33).

• The role of chemotherapy in LGA remains to be established. Because of concerns with regard to the morbidity of treatment with radiotherapy, chemotherapy has in recent years been chosen in preference to radiotherapy in infants and young children less than 3 to 5 years of age. Complete responses to chemotherapy are few, but the frequency of partial responses, minor responses, and stable disease is high, and overall response rates that include stable disease range from 70 to 100% 34, 35, 36, 37, 38, 39, 40, 41, 42. Progression-free survival rates after intervals of 2 to 3 years range from 23 to 100%, likely reflecting variations in the patient populations being studied rather than any effect of treatment per se. In one large recently reported series of newly diagnosed patients with symptomatic and/or progressive disease, progression-free survival was 75% at 2 years (42). In this series, only age at treatment correlated significantly with progression-free survival, younger children faring better than those older than 5 years, although in another series patients less than 5 years of age with chiasmatic tumors did very poorly with chemotherapy, with a progression-free survival rate of only 23% at 2 years (41). In both series, as well as in the recently completed POG study, patients with neurofibromatosis appear to fare better in terms of progression-free survival. Interestingly, in Packer’s series (42) there was no correlation between the type of response to chemotherapy (CR + PR vs. other) and progression-free survival, which may cast some doubt on the effectiveness of treatment with chemotherapy and tend rather to confirm that by-and-large, LGA in children follow an indolent course that depends on the specific tumor type and that is possibly little influenced by cytotoxic treatment. However, the current interpretation of these data is that the use of chemotherapy permits delay or even avoidance of radiotherapy. In Packer’s series, 45 of 78 patients who completed chemotherapy remained free from progression after a median interval of 36 months (42). In Janss’ series of LGA of the hypothalamic/chiasmatic region with a longer median follow-up of 72 months, 72% of patients treated with chemotherapy eventually progressed. Nonetheless, radiotherapy was delayed a median of 4 3/12 years to beyond age 5 years in more than 70% of the patients (41). The excellent (>90%) overall survival at 5 years in these studies as well as in the POG study has been interpreted as a consequence of successful initial and salvage therapy 41, 42. Again, however, longer follow up is necessary in this patient population. Studies currently underway in Europe (Societé Internationale d’Oncologie Pédiatrique, SIOP) and in North America (POG and CCG) in which the indication for starting treatment with carboplatin-based chemotherapy is clearly defined as patients with inoperable, symptomatic, and/or progressive disease will hopefully help elucidate the value of chemotherapy in children with LGA.

• Recent developments in radiotherapy suggest the need for reassessment of the risks of such treatment, and therefore, of its role in the management of LGA in children. Modern imaging techniques, functional imaging with MR and PET, computerized 3D treatment planning, the use of beam-modifying devices, and customized blocking have all contributed to greatly improved radiotherapy treatment delivery. In addition, there are now a number of alternatives to conventional external beam irradiation, including stereotactic irradiation, and brachytherapy. Such developments, in turn, have mandated more careful attention to the designation of target volume, and as well have permitted consideration of the use of higher doses of radiation.

In the past, some groups recommended the use of whole-brain radiotherapy for patients with LGA, even though the evidence for the need for such treatment was, at best, tenuous (43). With current imaging modalities and with better appreciation of what is and, more importantly, what is not, a low-grade astrocytoma, there is no justification for such an approach. Analysis of failure patterns, even after the limited fields used in conformal treatment (44), shows that almost all failures are local.

By extrapolation from the surgical literature, specifically the excellent tumor control achieved after complete and even subtotal resections, it seems clear that margins until now considered standard practice [e.g., 2–3 cm (19)] are overly generous, particularly for the well-circumscribed pilocytic tumors; margins of 0.5 cm around tumor as demonstrated on T2-weighted MR are probably adequate for such tumors. More generous margins of 1–1.5 cm are more appropriate for the more infiltrative fibrillary tumors.

Moreover, the fact that patients who have undergone grossly complete resection, inevitably with close resection margins, fare extremely well with disease-free survival rates close to 100%, suggests that the radiotherapy target volume need include only the residual disease, as demonstrated on postoperative MR. The resulting treatment volume will be considerably smaller than one based on preoperative imaging, as would until now have constituted standard practice.

Finally, there is the question as to whether it is necessary to include in the target volume the whole of a cyst. There are data in the surgical literature that demonstrate that removal of a nonenhancing cyst “wall,” in reality only compressed glial tissue, does not improve outcome 18, 45. In fact, morbidity is greater, and that would also be the case for radiotherapy, because the cyst is usually relatively large compared with the size of the solid component of the tumor. Thus, it seems reasonable in this situation to treat only the solid mass. On the other hand, a cyst wall that enhances with contrast material, a rare finding in LGA, is likely to represent tumor and should be included in the radiotherapy treatment volume.

Evidence for a dose–response correlation in LGA is scant (43). In some series, patients treated with higher doses fare better than those treated with lower doses; in others, the reverse is the case. It is noteworthy that a recent European Organisation for Research and Treatment in Cancer (EORTC) study that randomized adult patients with LGA between low dose (45 Gy) and high dose (59.4 Gy) radiotherapy failed to demonstrate any advantage for the higher dose (46). In spite of this, the tendency using newer, more conformal, radiotherapy techniques is to use a higher dose than the 54 Gy given in 30 fractions over 6 weeks considered standard therapy for children greater than 3 years of age. Although it may be feasible, because of a lower risk of injury to surrounding normal tissues using such techniques, to consider using higher doses, it remains to be seen whether this will result in improved tumor control.

The radiotherapy technique to be utilized is that which spares the greatest volume of normal tissue. With current imaging modalities, CT-simulation, image fusion and 3D planning, as well as the availability of radiotherapy treatment equipment that allows the use of multiple, sometimes noncoplanar field arrangements, customized field shaping, and intensity modulation of the radiation beam, conformity of the treated volume to the target is easier to achieve hitherto possible.

Nonetheless, for small tumors (less than 3 to 4 cm) stereotactic external beam irradiation may give even better dose distributions and preliminary results of single fraction treatment using the Gamma Knife (47) or fractionated treatment using linear accelerator based techniques 33, 48 are promising, even though the use of one or a relatively small number of treatment fractions may not be optimal for slowly growing tumors, particularly those comprised of tumor cells embedded in, rather than displacing normal brain, as is the case for LGA, particularly the fibrillary type. However, such treatment may be of considerable interest for part or even all of the treatment for pilocytic astrocytomas, which are often relatively small even at time of progression.

Brachytherapy has been used with some success, particularly in European centers 49, 50, 51, 52. Again, the principal limitation is the size of the target volume and brachytherapy series necessarily select for smaller tumors, with the result that there is little evidence at the present time that brachytherapy is a better treatment option than well-planned and well-executed external beam radiotherapy.

Radioactive solutions such as P-32, Y-90, Au-198 and Re-186 may be useful in cystic LGA (53). Often in these tumors, symptoms relate more to the cyst than to the solid component of tumor, which is frequently relatively small. Simple aspiration with or without placement of an internal drain will usually alleviate symptoms for protracted periods of time. However, in some cases, particularly those in which the cyst wall enhances and is felt to be biologically active, control of the cyst fluid may be difficult, and the use of radioactive solutions should be considered.