Stroke & Vascular Disorders
Neoplastic and infectious aneurysms
In this article, the author reviews current knowledge about intracranial aneurysms due to infectious and neoplastic causes. Direct mural injury or invasion
Jul. 21, 2021
At vero eos et accusamus et iusto odio dignissimos ducimus qui blanditiis praesentium voluptatum deleniti atque corrupti quos dolores et quas.
This article includes discussion of brainstem gliomas in adults, diffuse midline glioma, diffuse midline glioma with histone mutation, H3 K27M mutant glioma, high-grade midline glioma with H3 K27M mutation, medullary glioma, midbrain glioma, pontine glioma, cystic brainstem glioma, diffuse brainstem glioma, exophytic brainstem glioma, and focal brainstem glioma. The foregoing terms may include synonyms, similar disorders, variations in usage, and abbreviations.
In this article, the author discusses every aspect of brainstem gliomas, including pathology, molecular biology, clinical presentation, neuroimaging and diagnostic evaluation, differential diagnosis, prognosis and complications, surgical considerations, radiotherapy, and chemotherapy. In addition, a detailed list of references is included that contains over 100 articles. Although the article focuses on adult patients, relevant information from the pediatric literature is also included. In this latest update, the author summarizes several new reports on the molecular biology of brainstem gliomas as well as new approaches to therapy.
• Brainstem gliomas are much less common in adults than in children.
• Brainstem gliomas are a very heterogeneous group of tumors, but are often quite aggressive and difficult to treat, with relatively short survival times.
• A subgroup of brainstem gliomas harbor mutations in genes encoding components of the histone complex.
• Surgical resection of brainstem gliomas can be performed in selective focal tumors. For more diffuse tumors, resection is impossible, and a biopsy is preferred. In many cases of diffuse pontine tumors, MRI can be diagnostic.
• Radiotherapy is always necessary for high-grade tumors, but may not be needed for fully resected low-grade tumors. In general, radiotherapy is required for most brainstem gliomas.
• Chemotherapy has minimal activity against most high-grade brainstem gliomas, but it may be helpful in less aggressive tumor types.
The earliest descriptions of brainstem tumors in the literature were made in the 1930s (34; 70; 85; 15). In a review of 1737 verified brain tumors operated on by Harvey Cushing, Buckley found 25 located within the pons, the majority of which were astrocytomas (34). Gibbs noted that brainstem tumors were most common in children and estimated that the comparative incidence in adults was only 10% (70). A review by Hare and Woolf of 432 pediatric brain tumors seen at the Neurological Institute of New York included only 7 tumors of the brainstem (85). In a study evaluating diffuse pontine gliomas in children, Bailey was 1 of the first authors to draw attention to the high frequency of misdiagnosis in this population (15). Guillain and colleagues and Globus and colleagues were the first to document in detail brainstem tumors affecting the course of the aqueduct of Sylvius (79; 74). In a later review, Guillain was 1 of the first to document the behavior of exophytic brainstem tumors, in a review of 250 intracranial neoplasms (78). The first author to study a population of adult patients with brainstem gliomas was White, who evaluated a series of 44 patients accumulated over 31 years at the Neurological Institute in New York (233). The mean age of the patients was 42 years (range 17 to 68 years); most of those studied pathologically had astrocytic tumors of various grades. The clinical presentation and course of patients in this adult brainstem glioma series was similar to a comparable series of pediatric patients studied earlier at the Neurological Institute.
The clinical evolution and neurologic findings of adult patients with brainstem gliomas are similar to those seen in children and adolescents (233; 223; 142; 64; 205; 49). The length of time from onset of symptoms to diagnosis ranges from 2 to 10 months in most cases (median 4 to 5 months) and depends on the location of the tumor within the brainstem, the severity of initial symptoms, and the rapidity of progression of disease. In general, the most common initial neurologic complaints consist of gait abnormalities, visual disturbances, weakness, and headache (233; 223; 76; 166; 142; 203; 205; 80). Gait disturbance occurs in 40% to 80% of patients and is caused by weakness or spasticity from pyramidal tract dysfunction, ataxia, or a combination of these factors. In the study by White, 77% of an adult cohort of 44 patients complained of gait disturbance, and in the study by Tokuriki and colleagues, 42% of adult patients had a similar complaint (233; 223). Visual disturbances, which develop in 20% to 70% of patients, usually manifest as diplopia (horizontal more common than vertical). In the study by White, diplopia was the second most common complaint (233). Headache and focal weakness develop in 15% to 55% of adult patients (233; 223; 76). The headaches can be caused by several mechanisms, including expansion of the brainstem with traction on surrounding meninges and varying degrees of hydrocephalus. These headaches often occur in the suboccipital region, but their location can vary and may extend into the upper cervical area. In some patients, there may be a periorbital component of headache pain. Focal weakness can involve the face as well as 1 or more limbs and is usually asymmetric. Other common symptoms that developed in 15% to 40% of patients included dysarthria, focal numbness of the face or body, dysphagia, nausea and emesis, hearing loss, vertigo, tinnitus, and personality changes (233; 223; 203; 205). Dysarthria developed in 40% of patients in the study by White and was usually caused by disturbance of cerebellar function or corticobulbar pathways. Focal numbness was described by 35% of patients and ranged from hemisensory loss to numbness in 1 section of a limb. Dysphagia was a complaint in 10% to 30% of patients. The level of swallowing dysfunction can vary from slight difficulty handling liquids or solids to gross aspiration. Similar to other brain tumor patients, this group may also underestimate their level of swallowing difficulty (157). Nausea, emesis, and other related symptoms such as gastroesophageal reflux were present in 10% to 30% of patients and, in rare cases, can be the sole presenting complaint (233; 237; 223; 136; 61; 76; 138). These symptoms are probably caused by tumor-related compression of the vomiting center in the floor of the fourth ventricle, nucleus ambiguus, and dorsal vagal motor nucleus (61; 138). It is not uncommon for patients with this type of "gastrointestinal" presentation to be extensively worked up for gastrointestinal problems (eg, barium swallow, upper gastrointestinal series, endoscopy), often delaying the necessary neurologic evaluation and diagnosis. Treatment of the tumor may result in improvement of these "gastrointestinal" symptoms (237; 61). Alterations in personality are well documented in children with brainstem gliomas and can occur in adults as well; they were present in 15% to 18% of the patients studied by White (233; 166; 142).
At the time of diagnosis, the most common neurologic signs of brainstem gliomas in adults are pyramidal tract and cerebellar dysfunction, cranial nerve palsies, nystagmus, and gait difficulty (233; 223; 76; 166; 142; 203; 205; 80). In 65% to 75% of patients, pyramidal tract damage is noted, with contralateral weakness in an upper motor neuron pattern (233; 223). Over half of these cases will have hemiparesis or hemiplegia; monoparesis and tetraparesis are much less common. Two thirds of all patients have associated extensor plantar responses and asymmetric reflexes. Cerebellar involvement manifests as gait ataxia, limb dysmetria, or dysarthria in 65% to 80% of patients, most often caused by damage to corticopontocerebellar pathways or the cerebellar peduncles (233; 223). Cerebellar dysfunction contributes to the high incidence of nystagmus in this population and is noted in approximately 80% of all patients (149). The nystagmus is usually horizontal (65% to 70%) and gaze-evoked, less often during primary position. Infrequently, patients have vertical or rotary nystagmus.
Cranial nerve signs occur in the majority of patients (233; 223). In general, abnormalities of the lower cranial nerves (cranial nerves VIII through XII) are more common in adults than children and are usually caused by medullary tumors (223). The most frequent cranial nerve sign is palsy of cranial nerve VII, manifested by facial weakness in 50% to 80% of patients. Facial weakness is usually unilateral, but can be bilateral (5% to 7% of cases). Dysfunction of cranial nerves V (sensory), VI, IX, and X are next most common, each occurring in 45% to 55% of patients. Facial sensory loss is ipsilateral and variable in severity and can involve any or all of the trigeminal divisions depending on the region of tumor growth. Bulbar dysfunction caused by palsy of cranial nerves IX and X can range from mild palatal weakness to frank dysphagia, dysarthria, and aspiration. Decreased hearing is found in 40% to 45% of patients and can be caused by disturbance of the cochlear nuclei, trapezoid body, or other structures along the cranial nerve VIII pathways. Dysfunction of cranial nerves III, V (motor), XI, or XII can be noted in 20% to 30% of patients, whereas palsy of cranial nerve IV is rare. Other neuro-ophthalmological cranial nerve signs include papilledema, gaze paralysis (mainly horizontal), and anisocoria, each noted in approximately 25% of patients in the study by White (233). Disturbances of sensation are found in 35% to 55% of patients and usually present as a hemisensory deficit.
Unusual or rare neurologic signs and symptoms that may seem unrelated to a brainstem neoplasm can develop, such as isolated inability to sneeze, unilateral astereognosis, symptomatic cataplexy, isolated knee pain, and pathological emotions (eg, laughter, lability due to pseudobulbar palsy) (56; 143; 197; 44). Uncommon neuro-ophthalmologic abnormalities can also occur, including relative afferent pupillary defects, upbeat nystagmus, fatigable ptosis, and false-positive edrophonium testing for cranial nerve palsies (72; 149; 46; 114; 181). Facial myokymia and hemifacial spasm, which typically occur secondary to benign pathology outside the CNS, can occasionally be caused by brainstem gliomas, especially those arising in the pons (149; 232; 116; 83). In some cases, the facial manifestations may be the presenting feature of the tumor and can persist for many years before the diagnosis is confirmed.
Rarely, brainstem gliomas can affect regions within the medulla or pons that are involved in control of respiration and blood pressure (192; 98; 222; 226; 208; 234; 69). Apnea, apneusis, hypoventilation, and hyperventilation may occur when a tumor damages the lower brainstem pathways and nuclear regions responsible for respiratory control (192; 222; 226; 208; 234). Apneusis is characterized by marked prolongation of inspiration and can lead to life-threatening sequelae such as bradycardia, reduced oxygen saturation, and hypercapnia (234). Treatment with buspirone, a 5-HT1A serotonergic receptor agonist, can convert apneustic breathing patterns to normal (234). The hyperventilation responded to an oral continuous release preparation of morphine. Symptoms related to blood pressure dysregulation, especially orthostatic hypotension and secondary cardiac syncope, can also develop due to abnormalities of central autonomic control and disturbance of the afferent arc of the baroreceptor reflex (98). Abnormalities of these vital structures are conjectured to be responsible for the rare adult patients who present with sudden death from brainstem tumors (47).
The overall prognosis for survival and intact neurologic function in adults with brainstem gliomas is rather poor, but prognosis varies somewhat depending on the histology, location, and type of tumor (ie, diffuse, focal, exophytic, cystic) and its molecular characteristics (203; 58; 134). It is anticipated that our understanding of prognosis will continue to be refined as our understanding of the molecular subclassification of these tumors progresses. Reports of survival from the time of onset vary widely among authors, ranging from a mean of 19.2 months (median 12.5 months, 13 adult cases) in the group described by Tokuriki and colleagues, to a mean of 27.7 months (30 adult cases) in the cohort analyzed by White (233; 223; 166; 142). Data from the radiation oncology literature are similar, with median survivals ranging from 7.6 to 16.1 months (average median survival 13.6 months) in a collective cohort of 85 adult patients (51; 207; 81; 177). In the study by Guiney and colleagues, tumors located in the pons had a significantly adverse effect (p = 0.035) on survival when compared to other brainstem locations (81). Similar results have been noted by Schild and colleagues with 5-year survival of 54% with lesions outside the pons and 21% for those with lesions within the pons (p=0.04) (203). However, some authors have not found significant survival differences based on tumor location (51). In most of the studies from the radiation oncology literature, the percentage of patients surviving at 2 years was between 40% and 50% (113; 207). In a group of adult patients with pontine gliomas, the 3-year survival was 33% in a study by Guiney and colleagues (81). In contrast, a cohort of patients with pontine gliomas reported by Grigsby and colleagues had 5-year survivals of 46.6% (77). In studies that were able to compare survival between groups of adult and pediatric brainstem glioma patients who had received radiation therapy, results were comparable (113; 77; 81). In contrast, in a study by Landolfi and colleagues of 19 adult patients (median age 40 years) with brainstem gliomas as diagnosed by MRI scan, they noted a median survival of 54 months and a 5-year survival of 45% (121). The authors concluded that brainstem tumors may be less aggressive in adults than in children, even when located within the pontine parenchyma. Similar conclusions have been made by Guillamo and colleagues, who suggest that diffuse tumors may be less aggressive and of lower grade in adults than in children (80). In a review of 48 adult patients with brainstem gliomas, 46% of the cohort had a diffuse presentation on MRI. However, of the biopsied patients, 82% had benign histology with a median survival of 7.3 years. In the review of 143 adult brainstem glioma patients at MD Anderson Cancer Center, the median overall survival of the entire cohort was 32.1 months, which was similar to previous reports (219).
Numerous authors have evaluated brainstem glioma patients using univariate and multivariate analyses to determine both positive and negative prognostic factors. The vast majority of the data are derived from pediatric brainstem glioma patients (06; 40; 16; 166; 57). Fewer studies have included adult patients for prognostic evaluation (51; 76; 77; 206; 207; 81; 177; 203; 204; 80). In most of the studies, race, gender, extent of surgical resection, and age did not reach significance in univariate or multivariate analyses. The location of the tumor was a significant factor in several studies. Patients with tumors located in the pons often had shorter survival times (p = 0.035), whereas Shibamoto and colleagues noted that patients with midbrain tumors often survived longer (p < 0.05) (76; 206; 81; 203). The histological grade of the tumor was significant (p < 0.05) in a few studies, such that patients with microscopically evident malignant features or the diagnosis of glioblastoma had a poorer prognosis and shorter survival (77; 206). One of the most consistently important prognostic factors was duration of symptoms prior to diagnosis. In many studies, patients with longer symptom duration had longer survival times; conversely, those patients with shorter symptom duration had shorter survival times (51; 77; 166; 81; 177; 58). The type of tumor (ie, diffuse vs. focal) had prognostic implications in several reports; patients with diffuse tumors had significantly shorter survival and time to tumor progression (p = 0.012 and p = 0.005, respectively) (51; 177; 57; 203). Schulz-Ertner and colleagues evaluated 26 adults in their analysis of prognostic factors in a series of 41 brainstem glioma patients (204). The only significant prognostic factors for survival in the univariate and multivariate analyses were the presence of clinical or radiological response 6 weeks after treatment. A few authors have reported that patients with neurofibromatosis type 1 and brainstem gliomas have an improved prognosis over those tumor patients without neurofibromatosis type 1 (150; 180; 153; 175; 20). In a review of 21 neurofibromatosis type 1 patients with brainstem tumors, only 10 had radiographic or clinical progression during a 9-year follow-up period (175). In 4 of these patients specific therapy was required; 1 required biopsy, 3 required excision, and 2 required radiotherapy. All 4 lesions were histologically low-grade and remained stable after treatment. Bilaniuk and colleagues studied 25 patients with neurofibromatosis type 1 that had brainstem tumors (diffuse in 12 of 25 patients) (20). MRI evidence of tumor growth occurred in 8 patients (32%), all with diffuse tumors. Two patients with diffuse tumors died of progressive disease. The authors concluded that although patients in their series had a better overall prognosis than brainstem glioma patients without neurofibromatosis type 1, the tumors did not always behave in a benign, indolent manner (especially diffuse lesions) and were more dangerous than other tumor types found in patients with neurofibromatosis type 1. Some authors recommend using pontine midsagittal diameters in combination with magnetic resonance spectroscopy to differentiate “benign” pontine enlargement associated with neurofibromatosis type 1 from the more malignant diffuse brainstem glioma (29). Patients with neurofibromatosis type 1 and pontine enlargement had lesions that were significantly smaller than diffuse brainstem gliomas. In addition, the N-acetylaspartate peaks were consistently preserved in the neurofibromatosis type 1 patients without diffuse tumors, suggesting retained neuronal elements within the brainstem. In a study of prognostic factors in a large adult cohort, univariate analysis suggested that age of onset less than 40 years, duration of symptoms greater than 3 months before diagnosis, Karnofsky performance status greater than 70, low-grade histology, and absence of "necrosis" on MRI were all favorable prognostic variables (80). On multivariate analysis, only duration of symptoms, histological grade of the tumor, and presence of "necrosis" on MRI were significant and independent prognostic variables.
A study by Kesari and colleagues evaluated prognostic variables in a cohort of 101 adult patients with brainstem glioma (110). On multivariate analysis, ethnicity, tumor location, clinical grade of the tumor, and patient age at diagnosis all had a significant effect on survival. Non-Caucasian patients had a hazard ratio 4.19 times larger than Caucasian patients (p = 0.0004). Patients with diffuse pontine tumors had a hazard ratio 3.58 times larger than patients with tumors of the cervicomedullary junction (p = 0.0006). The hazard ratio for patients with clinically high-grade tumors was 2.81 times larger than those with low-grade tumors (p = 0.0017). Patients with an older age at diagnosis had a hazard ratio 1.34 times larger than younger patients (p = 0.0032). In general, adults with brainstem gliomas had a better prognosis than children with similar tumors. The overall survival for all patients at 5 and 10 years was 58% and 41%, respectively.
A study by Dellaretti and colleagues reviewed the clinical data from 100 adult and pediatric patients with diffuse brainstem gliomas to clarify prognostic factors (45). The mean age of the adult cohort (N = 63) was 41 years, whereas the mean age for the pediatric cohort (N = 37) was 6.9 years. The mean overall survival was 57 months in patients with a diffuse low-grade tumor, but only 13.8 months for patients with diffuse high-grade tumors (p < 0.001). For patients with non-enhancing tumors on MRI, the mean survival was 54.2 months, but only 21.7 months for enhancing tumors (p < 0.001). The Kaplan-Meier median survival curves were similar between the adult and pediatric groups, with a trend for longer survival in the adult cohort (25 versus 16 months; p > 0.05). On multivariate analysis using Cox proportional hazards regression, histological grade was the only significant prognostic factor (p < 0.001).
Several authors have studied prognostic factors in relation to information obtained from CT or MRI scans. Shibamoto and colleagues evaluated the correlation between prognostic factors and CT features (206). They found a significant association between estimated tumor volume and survival. Patients with tumors less than 6 cm3 survived longer than those with tumors between 6 and 20 cm3 (p less than 0.05) and larger than 20 cm3 (p less than 0.005). Furthermore, patients with tumors demonstrating diffuse contrast enhancement survived longer than patients with tumors that had ring enhancement (p less than 0.05). In an MRI study by Barkovich and colleagues, a significant (p = 0.02) survival advantage was evident for midbrain tumors (16). The 2-year survival for patients with midbrain tumors was 100%, whereas for patients with medullary and pontine tumors it was 50% and 20%, respectively. It was further noted that patients in whom tumors produced only limited enlargement of the brainstem, or were focal with limited infiltration, had significantly better survival than patients with more diffuse tumors. This is in contrast to the study by Nelson and colleagues, who noted a significant correlation (p=0.005) between the onset of necrosis, as imaged by CT or MRI, and subsequent survival (156). Over one third of patients with pontine tumors had evidence of necrosis at diagnosis and had a median survival of 6.5 months. For those patients who developed necrosis after diagnosis, the median survival was 9.0 months. Similar results regarding the presence of necrosis have been reported by others (80). In an MRI study evaluating the impact of gadolinium enhancement on outcome, no correlation was noted overall or within specific subgroups (eg, pons, midbrain) (57). However, only a small number of patients were evaluated with both MRI and histological data. These results are in contrast to data from Lesniak and colleagues, who studied 89 patients with brainstem gliomas and noted that enhancement within the tumor imparted a positive prognosis (128). The majority of these tumors were low-grade and responded well to aggressive surgical resection. A review of MRI findings in 39 cases of diffuse intrinsic pontine glioma evaluated the prognostic significance of several pretreatment and posttreatment imaging features, including rostral or caudal tumor extension, metastases, basilar artery encasement, necrosis, intratumoral hemorrhage, enhancement, hydrocephalus, and dorsal exophytic components (87). None of the MRI parameters had any prognostic significance. In a similar study, Liu and colleagues attempted to correlate MRI features at presentation with response to radiotherapy (131). Thirty patients with brainstem glioma were analyzed; 23 were rated as responders to treatment, whereas 7 were classified as nonresponders. There were no statistically significant differences in MRI features between the groups. However, necrosis and enhancement were more common in responders, whereas tumor surrounding the basilar artery, compression of the fourth ventricle, and extension into the cerebellum were more common in nonresponders.
A collaborative report from the International and European Society for Pediatric Oncology DIPG Registries reviewed the clinical, radiological, pathologic, and molecular aspects of short-term survivors and long-term survivors of diffuse intrinsic pontine glioma (95). They reviewed the data from 1130 pediatric and young adult patients with radiologically confirmed diffuse intrinsic pontine glioma; the median survival for the entire cohort was 11.0 months. The long-term survivor cohort (N = 101, [10%], survival > 2 years) more commonly presented at an age younger than 3 years or older than 10 years of age (p < 0.001) and had longer symptom duration (p < 0.001). In contrast, the short-term survivor group more often presented with cranial nerve palsy (p = 0.008), ring enhancement on MRI (p = 0.007), necrosis (p = 0.009), and extrapontine extension (p = 0.04). In addition, the long-term survivors more often received systemic therapy at diagnosis (p = 0.005) and had the presence of HIST1H3B mutations (p = 0.002).
Complications of brainstem gliomas vary depending on the size, location, and type of tumor. The most common complication is hydrocephalus, which occurs in 25% to 30% of patients as seen on CT or MRI (115; 21; 23; 80). Hydrocephalus is especially common in tumors of the midbrain, midbrain tectum, and cerebral aqueduct, where the incidence at diagnosis approaches 100% (94; 23; 213). Leptomeningeal tumor spread is a complication that occurs in 10% to 50% of patients antemortem, but is often detected more frequently in autopsy studies (109; 139; 164; 166; 209; 75; 206; 142; 48; 80). The symptoms may be present at diagnosis but more often develop during local tumor progression. Rarely, brainstem tumors can spread outside the CNS, causing extraneural metastases (109; 39; 194; 160). Extraneural metastases usually develop in the context of tumors that have already extended into the leptomeninges. Common sites for extraneural metastases from astroglial brainstem tumors include the lung, pleura, and lymph nodes (160). In patients with leptomeningeal tumor and a shunting device, spread to the intraperitoneal cavity may occur and cause increased abdominal girth, ascites, and abdominal pain (160). Although rare (ie, less than 5%), seizures and other episodes of altered consciousness have been reported in brainstem glioma patients (233; 77; 166). Patients with dysphagia may develop pneumonia as a complication of tracheal aspiration of food or oral secretions (157). Symptoms of aspiration include coughing or choking during intake of solids or liquids. However, in some patients with dysphagia, symptoms can be minimal or nonexistent, such that aspiration remains clinically silent until pneumonia is well established (157).
A 31-year-old woman with an unremarkable past medical history initially presented with diplopia. A workup consisting of several noncontrast CT scans of the brain and a neurologic examination was essentially normal. The neurologic symptoms progressed over the next 5 months, prompting a second opinion and an enhanced MRI scan of the brain. The MRI demonstrated an enhancing mass in the left pons, compatible with brainstem glioma. A biopsy revealed fibrillary astrocytoma grade II. Postoperatively the patient did well except for loss of hearing on the left side. No further treatment was offered at this point. Her neurologic status deteriorated further with significant weakness and spasticity on the right side, gait instability, and dysphagia. A follow-up MRI scan 6 months later showed further growth of the tumor. A second surgery was then performed that removed tissue that was of higher grade-fibrillary astrocytoma grade III. After recovery from surgery, she underwent 6 weeks of external beam radiation therapy. Her function improved during radiation, with reduction of right-sided weakness, dysphagia, and gait imbalance. Six months later, she noted progressive right arm weakness and spasticity, headache, and gait difficulty. During the next 7 weeks, left-sided weakness and spasticity, dysarthria, and more severe dysphagia developed. Another MRI scan demonstrated further progression and growth of the brainstem tumor.
The patient was then referred for chemotherapy evaluation. Her baseline neurologic examination was remarkable for normal mentation, dysarthric speech, paralysis of abduction in both eyes, adduction in the right eye, diminished corneal reflex in the left eye, lower motor neuron facial paralysis on the left, reduced hearing and facial sensation on the left, poor gag response bilaterally, spasticity of all 4 limbs right greater than left, 3+/5 right hemiparesis, hyperactive reflexes right greater than left, right-sided extensor plantar response, exaggerated Hoffman signs bilaterally, and severe gait ataxia. She was placed into the intraarterial carboplatin (200 mg/m2 X 2 days, q3 to 4 weeks) and intravenous etoposide (100 mg/m2 X 2 days, q3 to 4 weeks) protocol. After 6 cycles of chemotherapy the patient was improved neurologically and had a partial response (ie, 50% to 99% reduction) by MRI. She remained stable through 4 more cycles of intraarterial chemotherapy, over a total of 9.5 months, before radiologic progression. Treatment was changed to intravenous BCNU (200 mg/m2, q6 to 8 weeks). The patient remained stable for 4 more months before further neurologic deterioration developed. She was then taken off all treatment and died 6 weeks later.
The underlying etiology of gliomas, including brainstem gliomas, is incompletely understood (43; 108). The tumor cells most often have features of an astrocytic lineage; however, oligodendroglial, ependymal, and neuronal lineage cells can also be seen.
On pathological examination, brainstem gliomas of adults are similar in appearance to brainstem tumors from children. However, the distribution of tumors is different in the 2 groups. In children, over 75% of brainstem gliomas arise within the pons, 20% occur in the medulla, and 10% develop in the midbrain (223; 166; 142). In adult cases, 56% of tumors develop in the pons, 30% arise from the medulla, and 12% occur in the midbrain (223; 229). On gross inspection, the brainstem is usually swollen or deformed. The swelling can be diffuse, most commonly within the pons (ie, classic "pontine hypertrophy"), or it can be a more focal, lobulated, or an asymmetric alteration of normal medullary, pontine, or midbrain architecture. Tumors that cause symmetric expansion of the brainstem are usually diffuse, infiltrative, fibrillary astrocytomas of varying grade (often malignant). Intrinsic focal tumors, or those with a predominantly exophytic growth pattern, are more commonly of lower grade and reduced infiltrative potential (223; 166; 142; 141; 64). The exophytic regions of tumor are typically located on the floor of the fourth ventricle, within the cerebellopontine angles, or extending from the anterior aspects of the pons or midbrain into the prepontine or perimesencephalic cisterns. Ventrally exophytic tumor can frequently encase and displace the basilar artery, usually without causing secondary arterial thrombosis, ischemia, or infarction. Cysts may be present in low- or high-grade brainstem gliomas in up to one third of cases. In malignant tumors, the cysts often contain hemorrhage and necrotic debris. Large cysts or regions of tumor that encroach on the ventricular system (eg, floor of third ventricle, fourth ventricle, cerebral aqueduct) may cause hydrocephalus (166; 142). In slowly growing focal tumors of the midbrain tectum, hydrocephalus may develop in an indolent fashion as the neoplasm slowly compresses the cerebral aqueduct (227; 213). On rare occasions, gross deposits of tumor can be seen at a distance from the primary site within the brainstem (164; 209; 142). Tumor deposits are most common at the base of the brain, along the brainstem, and on the surface of the spinal cord and cauda equina. Distant metastases tend to occur more frequently in high-grade tumors, but can develop in low-grade fibrillary or pilocytic astrocytomas.
At the microscopic level, the vast majority of brainstem tumors in adults and children are of astrocytic lineage and appear similar to astrocytic tumors in other CNS locations (233; 139; 223; 142; 203). Fibrillary astrocytomas of various grades, growing in a diffuse or focal pattern, are the most commonly diagnosed neoplasms. A small subgroup of patients has pilocytic astrocytomas, whereas rare histologic diagnoses include oligodendroglioma, ganglioglioma, ependymoblastoma, hemangioma, mixed glioma, and primitive neuroectodermal tumors (178; 68; 117; 10; 127; 58; 246). Initial antemortem histological evaluation discloses low-grade, well-differentiated pathology in approximately 60% of biopsy and resection specimens; malignant pathology, usually anaplastic astrocytoma or glioblastoma, is found in 40% (233; 11; 223; 206; 28; 182). In contrast, autopsy studies demonstrate a higher percentage of malignant pathology, approximately 60% to 90% (139; 75; 206). This discrepancy is explained by the selection bias in autopsy materials toward more aggressive tumors and the tendency of well-differentiated tumors to undergo malignant degeneration during clinical progression. Furthermore, autopsy studies suggest that tumor spread and grade of malignancy depend on the site of origin of the neoplasm (139). In a study of 25 autopsied brainstem glioma patients (41% over 20 years of age), Mantravadi and colleagues found that pontine tumors were more likely than midbrain or medullary tumors to invade adjacent regions of brain and to have malignant histology (71% of pontine tumors were either anaplastic astrocytoma or glioblastoma) (139). A review of 143 cases of adult brainstem glioma at MD Anderson Cancer Center noted that 28 were grade IV, 43 were grade III, and 15 were diffuse astrocytomas (grade II), whereas 11 were unspecified gliomas; in 46 cases the diagnosis was made radiographically (219). In over 89% of the cases, the tumor was considered diffuse.
Low-grade fibrillary brainstem astrocytomas can demonstrate either a diffuse or a focal growth pattern but appear similar histologically. They are classified as astrocytoma, grade I or II, using the older Kernohan or St. Anne-Mayo classification schemes, or as grade II diffuse astrocytoma IDH-mutant by the World Health Organization (134). The tumors show a mild increase in cellularity and are composed of cells with astrocytic morphology. Cellular and nuclear atypia and pleomorphism are minimal. Rosenthal fibers are densely eosinophilic rod-shaped structures that may be present in some tumors. Low-grade tumors generally stain strongly with glial fibrillary acidic protein. Malignant features such as mitoses, endothelial proliferation, and regions of necrosis are not present. The most recent iteration of the WHO classification includes the presence of either IDH-1 or IDH-2 mutations in low-grade (grade II) infiltrating astrocytomas as a defining feature.
The growth pattern of low-grade fibrillary brainstem astrocytomas is to insinuate cells from the growing edge of the tumor between and along adjacent nerve fiber fascicles and pial-limiting membranes, expanding in accordance with the surrounding anatomy down a "path of least resistance" (54; 141; 42). Destruction of tissue and infiltration into gray matter is minimal. Because of the orientation of the pial-limiting membranes surrounding the brainstem and the alignment of the internal fiber tracts, enlargement of slowly growing, low-grade tumors is more likely to be directed toward the floor of the fourth ventricle and cerebellopontine angle regions, where there is less resistance to expansion (54). This explains the high frequency of exophytic growth noted in these areas. Low-grade tumors are also less invasive and infiltrative than more malignant tumors because of a reduced capacity to attach to, enzymatically degrade, and migrate through localized tissue extracellular matrix proteins (141).
High-grade fibrillary brainstem astrocytomas usually grow in a diffuse fashion, but, on occasion, they may have a more focal or exophytic presentation (127). They are classified as either anaplastic astrocytoma or glioblastoma, which correspond to astrocytoma grades 3 or 4, respectively, of the Kernohan, St. Anne-Mayo, and World Health Organization classification schemes (42; 134). These tumors demonstrate medium to high cellularity, with reduced amounts of cytoplasm, extensive nuclear and cellular pleomorphism, endothelial proliferation, mitotic figures, and, in some cases, necrosis.
When tumor cells infiltrate into gray matter, they tend to cluster around neurons (ie, perineuronal satellitosis) and collect in subpial and subventricular zones. Both anaplastic astrocytoma and glioblastoma are histologically heterogeneous tumors, but this is especially so for glioblastoma , in which it is not uncommon to have low-grade regions of astrocytoma near areas containing numerous mitoses, endothelial proliferation, and necrosis. High-grade astrocytomas also stain with glial fibrillary acidic protein, but in general the staining is less intense and more irregular than in low-grade tumors. Work by Jallo and colleagues has investigated the development of a reliable animal model that reproduces the aggressive growth characteristics of malignant brainstem gliomas (102). The authors injected 9L gliosarcoma cells or F98 glioma cells into the anterior pons of Fischer rats, using stereotactic techniques. The growth of the tumors was very reliable and led to the onset of hemiparesis within a mean of 16.5 days in 100% of the animals. Histopathologically, the tumors were similar to high-grade brainstem gliomas in humans.
The growth pattern of high-grade brainstem astrocytomas is different from that of low-grade tumors. The extent of infiltration and invasiveness into adjacent normal brain is more pronounced. In addition, as these tumors enlarge, they more often damage or destroy the underlying normal brain parenchyma. Structural limitations to growth within the brainstem, such as the anterior and lateral pial-limiting membranes and various large nerve fiber bundles, are not effective against high-grade tumors and do not direct growth like low-grade neoplasms (54; 141). This indiscriminate growth pattern explains the reduced frequency of exophytic extension in high-grade tumors, whose tendency is to infiltrate axially within the brainstem, although exophytic growth can occur in rare cases (127). The invasive and infiltrative nature of high-grade tumors is related to their highly developed ability to interact with, degrade, and migrate through specific substrate molecules found within the extracellular matrix of surrounding normal brainstem parenchyma (141).
A newly defined entity, diffuse midline glioma, H3K27M-mutant, has been added to the most recent WHO classification system. They are defined molecularly by mutations within specific histone genes, H3F3A (encoding histone 3.3) and HIST1H3B (encoding histone 3.1), leading to the substitution of a lysine by a methionine at position 27. Although not exclusively located in the brainstem, this midline location is one neuroanatomic site where they can be found. This category includes the diffuse-intrinsic pontine gliomas (DIPGs). This class of tumors occurs predominantly in the pediatric population (134; 147).
Pilocytic astrocytomas typically arise in the dorsal pons, with subsequent growth rostrally through the brainstem, but can develop anywhere (58). They are usually focal, indolent, slow-growing neoplasms and often have an exophytic component (111). Pilocytic astrocytomas are designated astrocytoma, grade I by the World Health Organization (42). On microscopic evaluation they most often display a biphasic pattern of loose, microcystic areas, alternating with more dense regions composed of elongated, "hairlike" astrocytic cells.
Rosenthal fibers and granular bodies are often numerous in the dense regions of tumor. Aggressive features can be present, including cellular atypia, occasional mitoses, multinucleated cells, focal necrosis, and prominent vasculature. However, these elements do not indicate malignancy or alter the more favorable prognosis and improved survival of patients with this type of brainstem tumor. Some authors feel pilocytic astrocytomas of the brainstem should be classified separately from fibrillary astrocytomas in this region because of their vastly different clinical presentation and extended survival (58).
Gangliogliomas of the brainstem are uncommon tumors composed of an admixture of neoplastic neuronal (ie, ganglionic) and astrocytic cells (68; 117; 42; 250). The astrocytic portion of the tumor predominates and is usually classified as being fibrillary and of low grade. Mitoses and other aggressive features are generally absent. Gangliogliomas are typically slow growing and well circumscribed, without significant infiltration. In a literature review of 14 patients with brainstem gangliogliomas by Garcia and colleagues, 5 were older than 18 years of age (36%), and in 11 of the patients (78.5%) the medulla was the site of neoplastic origin (68). A Chinese study reports 7 patients with a mean age of 28.6 years; all of the tumors were classified as WHO grade I or II, and did well with aggressive surgical resection (250). Rarely, other types of mixed glioma can develop within the brainstem in adults (eg, mixed anaplastic oligodendroglioma or ependymoma) (127).
Techniques that can measure the biological potential of brainstem gliomas include Ki-67 and bromodeoxyuridine labeling studies, and functional brain imaging using SPECT or PET scans. Both labeling methods quantitatively assess the proliferating fraction and, therefore, the inherent aggressiveness of a tumor (ie, labeling index). Ki-67 reacts with cells in G1, S, and G2M phases, whereas bromodeoxyuridine labels only cells in S phase. For fibrillary astrocytomas, in general, Ki-67 and bromodeoxyuridine labeling indices increase with grade, whereas for all pilocytic tumors the indices tend to be low (100; 103). Several authors have evaluated brainstem gliomas using functional imaging methods to differentiate between tumor progression and radiation necrosis (33). Bruggers and colleagues studied a brainstem glioma patient with serial MRI and 18-fluorodeoxyglucose PET scans (33). The PET scans demonstrated increasingly hypermetabolic activity, which was helpful in differentiating between tumor progression and radiation necrosis when MRI scans remained stable. Autopsy revealed a diffuse, viable glioblastoma without evidence of radiation necrosis.
Cytogenetic and chromosomal studies of brainstem astrocytomas are uncommon, most likely due to the paucity of tissue available for analysis following surgery, which is usually a small biopsy. However, they are probably similar to astrocytomas of the same grade from other CNS locations (186; 105).
Work by Zhou and colleagues has evaluated the presence of B7-H3 (also known as CD-276) in the cell membranes of diffuse brainstem glioma tumor cells (252). B7-H3 is a membrane protein involved in host defenses that has been shown to be overexpressed in malignant neuroectodermal tumors. Monoclonal antibodies against B7-H3 have shown efficacy as salvage therapy in children with stage IV neuroblastoma. In a series of brain tumor specimens, all 9 of the diffuse brainstem glioma specimens showed immunoreactivity for B7-H3, whereas none of the WHO grade I tumors, and only 37.5% of the WHO grade II tumors demonstrated any reactivity. The authors concluded that B7-H3 was overexpressed in diffuse brainstem glioma, and that monoclonal antibody therapy directed at B7-H3 warrants further investigation.
Our understanding of the role of oncogenes and tumor suppressor genes in the origin of brainstem gliomas has evolved over time. The most recent impactful advances revolve around the role of histone mutations in these tumors. A historical perspective highlighting specific mutated genes or pathways will be provided in this article. A study by Louis and colleagues evaluated archival or autopsy tissue from 7 patients with brainstem glioblastoma (2 patients 16 years of age) for abnormalities related to epidermal growth factor receptor amplification, loss of heterozygosity of chromosomes 10 and 17, and mutations within the p53 gene using polymerase chain reaction-based techniques (135). None of the tumors demonstrated any amplification of the epidermal growth factor receptor gene. Four of 7 tumors (57%) had allelic loss of chromosome 10. Four tumors were also shown to have lost portions of the short arm of chromosome 17 (57%; all included the region containing the p53 gene, 17p13.1). The p53 gene was mutated in 5 of 7 tumors (71%). Four of these 5 cases were p53 genes on the remaining allele of tumors that had lost heterozygosity, whereas the fifth was from a tumor with both copies of 17p. Two of the p53 mutations caused base changes that resulted in amino acid substitutions (Arg Cys, Arg Leu), whereas a third resulted in a stop codon. The authors concluded that these results were consistent with previous data suggesting that a subset of glioblastoma that develops in younger patients is derived from lower-grade, diffuse, fibrillary astrocytomas (228). The lower-grade tumors and subgroup of glioblastoma are characterized by allelic loss of 17p, p53 mutations, and a lack of epidermal growth factor receptor amplification. These data are in agreement with Zhang and colleagues, who found mutations within the p53 gene in 8 of 13 pontine anaplastic astrocytomas (62%; 4 patients 16 years of age) (249). Of 15 total mutations detected, 11 were missense mutations that resulted in amino acid substitutions. Another report has demonstrated similar results using immunohistochemical methods, with p53 accumulation more frequently noted in the higher-grade tumors (14). A report by Lang and colleagues describes an adult patient with a brainstem anaplastic astrocytoma that did not show evidence for loss of heterozygosity of chromosomes 10 or 17p, mutations within the p53 gene, or amplification of the epidermal growth factor receptor gene (122). Molecular analysis of 7 pediatric patients with primitive neuroectodermal tumors of the brainstem did not disclose p53 mutations (246). This is similar to primitive neuroectodermal tumors of the cerebellum and cerebrum, which are also lacking in abnormalities of p53. A molecular analysis of 28 specimens of pediatric brainstem gliomas has detected a high incidence of amplification and overexpression of ERBB1 (ie, EGFR) (71). The degree of ERBB1 expression correlated closely with increasing tumor grade (p< 0.001) and was not related to p53 status. The percentage of EGFRvIII mutations in the amplified specimens could not be determined with the techniques used. The authors suggest that ERBB1 signaling is important for the pathogenesis of many pediatric brainstem gliomas and that molecular therapeutic agents designed to target this pathway (eg, ZD1839; Iressa) should be evaluated. Verification of ERBB1 amplification and expression status in brainstem gliomas in adults remains unclear. Using high-resolution single nucleotide polymorphism (SNP)-based DNA microarray analysis, a series of 11 samples from patients with diffuse pontine gliomas was analyzed (247). In 36% of the cohort, gains in copy number (range 4 to 18) were noted for platelet-derived growth factor receptor alpha (PDGFR-alpha). Expression of PDGFR-alpha was noted in all tumor samples. In 3 cases, low-level gains were present for poly(ADP-ribose) polymerase (PARP)-1. Pathway analysis also noted genes with loss of heterozygosity in the DNA repair pathway. The authors suggested that PDGFR-alpha and PARP-1 might be rational targets to consider for molecular therapy of brainstem gliomas.
Paugh and colleagues used single-nucleotide polymorphism arrays to evaluate genomic copy number abnormalities in a series of 43 diffuse pontine gliomas (170). They noted focal amplifications of genes within the receptor tyrosine kinase-Ras-phosphoinositide 3-kinase signaling pathway in 47% of the tumors, mostly involving PDGFR-A and Met. In 30% of the tumors, there were focal amplifications of cell-cycle regulatory genes controlling retinoblastoma protein phosphorylation, whereas concurrent amplification of genes from both pathways was present in 21% of the tumors. These amplification patterns differed from low-grade brainstem tumors and high-grade gliomas outside the brainstem. A clinicopathological review of 25 brainstem gliomas (10 diffuse) by a Japanese group noted that of the diffuse cohort, 4 were grade II, 4 were grade III, and 2 were grade IV (162). The MIB-1 labeling index varied from 0.8 to 38%. Additionally, p53 was positive in 80% of tumors, whereas MGMT was positive in 60%, and IDH1 was negative in 61.6% of tumors.
In an analysis of 17 diffuse intrinsic brainstem gliomas in adults, Reyes-Botero and coworkers screened for mutations of isocitrate dehydrogenase (IDH1), histone genes, p53, Ki67 labeling index, and MGMT promoter methylation and compared the results to a large series of adult supratentorial gliomas (188). The median survival of the cohort was 48.7 months (57 months for low grade versus 16 months for high grade; p < 0.01). IDH1 mutations were uncommon, noted in only 3 tumors (R132G, R132C, R132H). Mutations within histone genes were more common, in particular H3F3A and HIST1H3B. In addition, a higher frequency of chromosome arm 1q gain, 9q gain, 11q loss, and less frequent chromosome 7 gain and chromosome 10 loss, were noted. No EGFR amplification was noted in the cohort. New mutations have also been reported by several authors in the activin receptor gene ACVR1 (35; 217). ACVR1 encodes a type I activin receptor serine/threonine kinase. After the mutation, the gene becomes constitutively active, leading to SMAD phosphorylation and increased expression of downstream activin signaling targets ID1 and ID2. These mutations are similar to the mutations found in the germ line of patients with the childhood developmental disorder fibrodysplasia ossificans progressiva.
The role of the Wnt/beta-catenin pathway in the development of brainstem glioma is now under investigation. Data from Wu and colleagues reviewed the expression of Wnt-1, Wnt-2, beta-catenin, and C-myc in a series of 24 brainstem glioma, as well as normal brain controls (238). Wnt-1 did not have significant expression in normal brain, but was present to the same degree in low-grade and high-grade brainstem glioma. The expression of Wnt-2, beat-catenin, and C-myc was significantly higher in high-grade brainstem glioma in comparison to normal brain and low-grade brainstem glioma, and was also strongly correlated with the expression of Ki-67. In addition, Wnt-2 and C-myc were significantly associated with prognosis in brainstem glioma patients. Additional work from the same investigators also showed that the protein and mRNA levels of beta-catenin and N-cadherin were significantly associated with the pathological grade of brainstem glioma (239). In addition, higher expression of beta-catenin was associated with a shorter overall survival in brainstem glioma patients.
Another somatic mutation discovered by genome-wide analyses of brainstem gliomas is the PPM1D gene, which seems to function as a proto-oncogene (185). PPM1D (also called WIP1) demonstrated heterozygous mutations in 9% to 23% of diffuse intrinsic pontine gliomas. The gene dephosphorylates multiple DNA damage response proteins including p53, Chk2, ATM, and ϒ-H2Ax, thereby allowing reentry into the cell cycle. The PPM1D mutations in brainstem gliomas lead to a truncated form of the protein that has a longer half-life and more persistent impact on the cell cycle. Small molecule PPM1D inhibitors have been developed, and need to be used against brainstem gliomas in vitro and in animal models.
Using an implanted C6 glioma model of brainstem glioma in rats, Liu and coworkers attempted to discern differences in the growth patterns and biology of the juvenile and adult animals (132). Adult rats presented with focal neurologic symptoms and had more focal histology within the pons. In contrast, juvenile rats had a more rapid presentation and a more diffuse neurologic presentation, with ataxia, cranial nerve deficits, and incontinence. Histologically, the tumors were more diffuse and infiltrative throughout the entire brainstem. In addition, tumors in juvenile rats had a higher Ki-67 labeling index and a lower rate of apoptosis. The median survival was significantly higher in the adult rats (p < 0.05). The authors concluded that the immature rat brainstem may offer or create a more permissive cellular environment, allowing tumor cells to be more invasive.
Micro-RNA are small non-coding RNA molecules that are involved in the transcriptional control of gene function and have been implicated in the pathogenesis of cancer. They can function as oncogenes or tumor suppressor genes through their interactions with mRNA. In a study by Wang and colleagues, micro-RNA expression profiles in orthotopic models of brainstem glioma were analyzed (230). They noted that miR-20a and miR-106b were the most robustly upregulated mi-RNAs during the process of malignant progression in pediatric brainstem glioma.
The 2016 update to the WHO classification system incorporated a new category of infiltrating gliomas, high-grade midline gliomas with H3 K27M mutation. Patients with diffuse intrinsic pontine gliomas that harbor the H3 K27M mutation have been shown in more than 95% of samples to have a profound epigenetic change, related to impaired function of polycomb repressive complex 2 (PRC2) methyltransferase, resulting in global hypomethylation of H3K27 (147). In addition, 3 main molecular subgroups have been identified in these tumors -- H3 K27M, MYCN, and Silent -- by integrating DNA methylation patterns, whole genomic sequencing, and histopathological and clinical data. H3 K27M is the most prevalent and is characterized by the histone mutations, global DNA hypomethylation, and multiple concurrent mutations (eg, p53, PAX3, EGFR). The MYCN subgroup is characterized by high-grade histology, chromothripsis on chromosome 2p, and DNA hypermethylation. The Silent subgroup is characterized by a lower genomic mutational burden than the other 2 subgroups.
To obtain tumor tissue for molecular phenotyping, Pan and coworkers used CSF samples to analyze circulating tumor DNA (ctDNA) for genomic profiling, as an alternative to biopsy, in brainstem glioma patients (169). They obtained CSF samples in 57 patients with brainstem gliomas, including 23 with diffuse pontine tumors. At least one tumor-specific mutation was detected in over 82.5% of CSF ctDNA samples (47 of 57). Mutation detection using plasma ctDNA was less sensitive than sequencing the CSF ctDNA (38% versus 100%, respectively). The authors concluded that CSF sampling and ctDNA analysis was a reliable technique and may offer an alternative approach to stereotactic biopsy for molecular phenotyping.
Zhang and colleagues performed a molecular and immune analysis in a series of 62 patients with malignant brainstem tumors that had undergone tumor resection (251). IDH1 mutations were only noted in adults and occurred more frequently in large tumors (> 10 cm3; p = 0.01), and were mutually exclusive with H3F3A mutations. In contrast, H3F3A mutations were more common in children and teenagers (p = 0.013), those with low preoperative Karnofsky Performance Scale scores (p = 0.021), and higher-grade brainstem tumors (p = 0.038). MGMT promoter methylation was mainly noted in the IDH1 mutated tumors and was very rare in tumors with H3F3A mutations. PD-L1 staining was present in 59.7% of specimens. High intra-tumoral CD8+ T-cell density was less frequent in the H3F3A-mutated tumors than the H3F3A-wild-type tumors (p = 0.005). Patients with H3F3A-mutant tumors had a much worse prognosis in comparison to those with IDH1 mutant or H3F3A-IDH1 co-wild-type tumors (p = 0.001). The authors concluded that resectable malignant brainstem gliomas could be classified into 3 subtypes: H3F3A-mutant, IDH1 mutant, and H3F3A-IDH1 co-wild-type tumors.
Brainstem tumors comprise 2% of all histologically diagnosed primary brainstem tumors (163). In children, they make up 10% to 20% of pediatric primary brain tumors (157; 64; 49). Furthermore, approximately 70% to 75% of these neoplasms develop by 20 years of age (142). By comparison, the incidence in the adult population is estimated to be 9 to 10 times lower than that in children, roughly 0.5% to 2.0% (233; 223; 130; 157). In a study by Gibbs examining 1383 primary intracranial tumors in patients aged 16 to 70 years, only 1% were located within the pons or midbrain (70). Tokuriki and colleagues reviewed 3044 adult primary brain tumors over a 35-year period (223). Intrinsic tumors of the brainstem were found in 16 cases, an incidence of only 0.52%. However, in smaller series of brainstem glioma patients, the proportion of adults may be larger, ranging from 10% to 35% (233; 76). The vast majority of brainstem gliomas are sporadic, but they can be familial on rare occasions (202; 59).
There are no known measures to prevent the development of a brainstem glioma in de novo patients or in those at risk from associated disorders (eg, neurofibromatosis type 1).
The differential diagnosis of brainstem gliomas in adults consists of other mass lesions or disease processes that can present with a progressive brainstem syndrome and includes other neoplasms, parasitic cysts, viral and postviral brainstem encephalitis, infarction, tuberculoma, demyelinating syndromes, vascular malformations, abscesses, radionecrosis, and chronic meningitis of various etiologies (60; 63; 01; 210; 142; 182; 119). Neoplastic considerations include metastases to the brainstem or posterior fossa, as well as primary brain tumors of the posterior fossa or skull base such as meningioma, acoustic schwannoma, dermoid cyst, medulloblastoma, ependymoma, and epidermoid cyst. Infectious considerations include parasitic cysts from cysticercosis, which have a predilection for the fourth ventricular region, tuberculomas or pyogenic abscess formation of the posterior fossa, viral encephalitis involving the brainstem, and chronic meningitis with basilar exudate (eg, tuberculosis, Cryptococcus neoformans). Demyelinating syndromes refer mainly to multiple sclerosis, which often presents with multifocal brainstem signs and symptoms, and to postinfectious brainstem encephalitis, which is considered to be an autoimmune-mediated demyelinating process following viral exposure. In 1 case report, a 48-year-old woman demonstrated a progressive brainstem syndrome, along with encephalopathy and hallucinations (119). She was thought to have an acute demyelinating encephalopathy or collagen vascular disease, and she was treated with plasmapheresis, without benefit. At autopsy, a large brainstem glioma was noted, which involved the pons, medulla, and cerebellum. Cerebrovascular processes include brainstem infarction (although these tend to occur rather acutely), hemorrhage, and various vascular malformations (ie, arteriovenous malformation, capillary telangiectasia, cavernous angioma) that can become symptomatic because of mass effect or spontaneous hemorrhage.
Although CT and MRI will usually help to discriminate between brainstem glioma and other disease processes within the posterior fossa, they are not 100% sensitive or specific and may lead on occasion to false-positive or false-negative diagnoses (60; 63; 01; 182). In a study by Frank and colleagues, 17% of 38 patients with a preoperative diagnosis of a brainstem tumor had a nonneoplastic lesion at biopsy, usually a vascular abnormality or malformation (60). In 24 adult patients with expanding brainstem masses, 25% had nonneoplastic diagnoses following stereotactic biopsy (63). Rajshekhar and Chandy describe a series of 71 patients with brainstem lesions in which 19.4% had diagnoses other than brainstem glioma following biopsy (182). This group of nongliomatous lesions included nonspecific chronic inflammation (4 patients), granulomatous inflammation (2 patients), epidermoid cyst (2 patients), pyogenic abscess (2 patients), and encephalitis (1 patient). In a subgroup of 7 patients with the preoperative CT diagnosis of glioma in which benign lesions were found at biopsy, the diagnoses included encephalitis, epidermoid cyst, and tuberculoma. In the study by Abernathy and colleagues, 26 patients (20 over 18 years old) with pontine mass lesions were biopsied for suspected brainstem tumors (01). Ten of 26 patients (38.5%) had a benign diagnosis, including cryptococcal abscess, arteriovenous malformation, demyelination, infarction, and radionecrosis.
In patients with neurologic signs and symptoms suggestive of a brainstem glioma, neuroimaging with CT or MRI is the most critical diagnostic test. These studies should be performed with and without contrast media for the most accurate visualization of the mass, to assess the relationship of the mass to other posterior fossa anatomy, and to assist in differential diagnosis. The appearance of brainstem gliomas in adults on CT or MRI is similar to that in children. In general, both CT and MRI demonstrate enlargement and asymmetry within the brainstem, most commonly the pons, with minimal edema formation around the tumor (223; 206; 210; 166; 142; 141; 49).
However, MRI has been shown to be superior to CT for diagnosing brainstem gliomas, with a sensitivity and specificity for pathology approaching 100% (27; 99; 126; 168; 254; 179; 16; 210; 166; 142; 141; 23; 57). This improvement in diagnostic accuracy is due to several factors, including a lack of bone artifact in the posterior fossa, increased sensitivity to intratumoral water content, and the multiplanar capability of MRI, especially the midsagittal view. In 10% to 20% of cases, MRI will detect tumors that are not discernible by CT, especially small, focal tumors or those that occur within the midbrain. This is important for patients who have hydrocephalus on CT, but with no evidence on the scan for an intrinsic brainstem tumor. All of these patients should be studied with MRI because some of them will harbor neoplasms that might benefit from earlier diagnosis and treatment (179). The anatomical configuration within the brainstem, extent of infiltration, presence of an exophytic component or leptomeningeal seeding, and degree of enhancement are visualized in more detail with MRI (27; 99; 126; 168; 254; 48). Despite these advantages, MRI is similar to CT in its inability to consistently differentiate between neoplastic and benign lesions or to correlate the appearance of a given lesion with a specific pathologic diagnosis. In addition, data would suggest that even with MRI, there is significant interobserver variability in the measurement of diffuse brainstem gliomas. Because of this variability, the authors suggest that for clinical trial purposes, the measurements of brainstem gliomas should be made by a single neuroradiologist (89).
When using MRI for the initial assessment of brainstem gliomas, it is recommended that T1- and T2-weighted nonenhanced imaging, as well as T1-weighted gadolinium-enhanced imaging, be performed in both the axial and midsagittal planes (16; 166; 49).
Doing this will provide the most information on the size, extent, and other characteristics of the tumor necessary for treatment decisions such as extent of surgery and design of radiation ports, and for detailed comparison at follow-up. On nonenhanced T1-weighted images, brainstem gliomas generally demonstrate hypointense signal compared to normal white matter (99; 126; 168; 254; 16; 166; 142). On proton density and T2-weighted images, increased signal intensity is usually noted within the mass.
Because the degree of peritumoral edema is typically insignificant in these tumors, the extent of abnormal T2-weighted signal is believed to represent bulky tumor and any associated infiltration. Following administration of gadolinium, T1-weighted images show variable amounts of enhancement, with patterns similar to that seen on CT: negligible, ring-enhancing, partial/focal, and diffuse.
The exophytic components of tumors are usually found in the basilar subarachnoid spaces or cerebellopontine angles, and they frequently enhance in a diffuse fashion.
Disturbance or encasement of the basilar artery by exophytic tumor in the prepontine cistern is clearly seen on MRI. Additional features that can demonstrate enhancement and suggest more aggressive behavior are subpial extension, subependymal extension, ill-defined tumor margins, necrosis, and the presence of daughter nodules (142; 141). The presence of acute or old hemorrhage within tumor is easily determined by MRI, although it is uncommon at presentation, occurring in approximately 6.25% of cases as reported by Broniscer and colleagues (31). However, symptomatic hemorrhage is noted in nearly 20% of patients after diagnosis during the course of therapy. On rare occasions, MRI can demonstrate brainstem tumors infiltrating along or encasing the lower cranial nerves (245; 184). Some authors report the importance of this information in presurgical planning of exophytic brainstem tumors (245). A study correlating MRI and PET findings noted that diffuse pontine gliomas were always of high-grade pathology (ie, glioblastoma or anaplastic astrocytoma), with variable exophytic features (118). PET scans were hypermetabolic only in the tumors with glioblastoma multiforme pathology. However, some glioblastomas and all of the anaplastic astrocytomas were hypometabolic, as were the low-grade tumors. MRI techniques, such as diffusion tensor imaging (DTI), are also being applied to brainstem gliomas (91; 92). DTI allows for a more detailed and quantitative evaluation of the fiber tracts in the brainstem and shows the degree of involvement of these tracts by tumor. Preliminary data suggest that DTI demonstrates superior visualization and quantification of tumor involvement in the motor, sensory, and transverse pontine tracts in comparison to conventional MRI. In addition, DTI appears to be able to demonstrate axonal degeneration within the motor and sensory tracts of the brainstem in these patients (92).
Several authors have applied MRI as a tool for classification of brainstem tumors into groups with treatment-related and prognostic significance (52; 53; 16; 54; 57; 80; 243). Epstein and colleagues have classified brainstem tumors into 4 groups: diffuse, focal, cystic, and those that occur at the cervicomedullary junction (52; 53; 54). Diffuse tumors are most common and have the worst prognosis, generally arising within the pons and demonstrating a hypointense signal that usually does not enhance after contrast. Extension into the medulla and midbrain is often noted. Sagittal views typically demonstrate significant pontine hypertrophy.
Focal tumors generally measure less than 2.5 cm in diameter, have no associated edema, and enhance with contrast. They are more likely to have an exophytic component, pilocytic histology, and a better prognosis than diffuse tumors (111). Cystic tumors that contain an enhancing mural nodule have an excellent prognosis similar to that of cerebellar pilocytic astrocytomas. If the tumor has an enhancing solid component in addition to an enhancing cyst wall, it is usually malignant and carries a poor prognosis. Tumors of the cervicomedullary junction are often of low grade, rarely extend above the pontomedullary junction (though occasionally growing caudally into the cervical spinal cord), and have a favorable prognosis. They are well delineated on sagittal MRI. Barkovich and colleagues have characterized the MRI scans of 87 "high-risk" brainstem glioma patients with respect to tumor location of origin, focality, direction and extent of growth, degree of brainstem enlargement, degree of exophytic growth, hydrocephalus, hemorrhage, necrosis, and presence or absence of cyst (16). The authors found a significant difference in survival based on the primary location of the tumor (p = 0.02): no deaths in 6 cases of midbrain lesions, 50% survival at 2 years for medullary tumors, and only 20% survival at 2 years for pontine tumors. The degree of brainstem enlargement had an inverse relationship with survival (p = 0.05): no or mild versus moderate or severe. Similarly, degree of infiltration was inversely related to survival (p = 0.005). The presence of a tumor cyst denoted a more favorable outcome (p = 0.03). Yin and Zhang report a series of 150 patients (120 of which were adults) with brainstem glioma who were evaluated by MRI and then underwent subsequent treatment, including microneurosurgery (243). The accuracy of the MRI diagnosis was estimated to be 95.3%. MRI evidence of a focal lesion was associated with a more favorable histopathological diagnosis in intrinsic (p = 0.005) and exophytic (p = 0.001) brainstem patients. Of the patients with an intrinsic diffuse tumor, those without enhancement had more favorable pathological findings (p = 0.009).
In a study of 96 patients with histologically verified brainstem gliomas, Moharamzad and coworkers investigated the association between specific imaging findings and histopathological grading (ie, low-grade versus high-grade) (152). There were 39 children (mean age 9.38 years) and 57 adults (mean age 35 years). They were evaluated on a conventional 1.5T MR unit for T1-, T2-, and T1-weighted postcontrast images. Logistical regression analysis revealed that for adults the presence of necrosis (p = 0.001) and inhomogeneous contrast enhancement (p = 0.001) were significant predictors of high-grade pathology. For pediatric cases, only inhomogeneous enhancement (p = 0.002) was a significant predictor.
Some authors have begun to apply MRI technology to monitor magnetic resonance spectroscopy (MRS) of tumors, including brainstem gliomas. A report by Laprie and colleagues used a multivoxel version of this technique to monitor brainstem gliomas during and after treatment (124). A series of 24 MRS studies on 8 patients was performed, and the data suggested that longitudinal multivoxel MRS was feasible, reliable, and potentially superior to univoxel techniques for evaluating response to radiotherapy and other treatment modalities. Two patients were described that had undergone longitudinal MRS monitoring of their diffuse brainstem gliomas during therapy (218). Although there was a brief interval of clinical improvement after radiation therapy, the MRS signature demonstrated further elevation of the Cho/Cr and Cho/NAA ratios, more consistent with progression. This was confirmed at the next follow-up visit, when both patients were noted to have clinical deterioration. The authors suggest including MRS in the routine radiological evaluation of patients with brainstem gliomas.
Although CT is not as accurate as MRI, it can still discern the presence of a brainstem tumor over 90% of cases, especially if the lesion is within the pons and has an exophytic component (115; 21; 01; 142; 141; 205). With unenhanced CT, the tumor usually appears hypodense or isodense compared to normal surrounding brain.
Effacement of the basal cisterns, displacement and deformity of the fourth ventricle, and disturbance of the basilar artery are additional early signs of tumor. Tumor enhancement is variable, occurring in approximately 50% of tumors. Exophytic portions are most likely to demonstrate strong enhancement. Diffuse, infiltrative tumors are generally hypodense and most often show partial or negligible enhancement (21; 223; 206; 210). Focal tumors, with or without exophytic components, are more likely to demonstrate diffuse or ringlike enhancement. Hydrocephalus is noted in 25% to 30% of tumors and is especially common in tumors affecting the midbrain tectal region (115; 21; 23). Cystic changes are noted in 10% to 15% of cases and can be associated with low- or high-grade tumors (115; 21; 210). CT is limited by beam-hardening artifact in the posterior fossa, which reduces sensitivity for small, focal lesions and is often unable to properly gauge the extent of tumor infiltration within the brainstem (210; 166; 142; 141).
Several authors are now applying positron emission tomography (PET) technology to brainstem gliomas. One group used 18F-fluoro-ethyl-tyrosine PET to analyze progression-free survival in 16 adult patients (04). Progressive tumors had higher SUVmax (p = 0.003) and TBRmax (p = 0.001) in comparison to stable tumors. The TBRmax was predictive of progression-free survival: it was longer in patients with a TBRmax less than 2.0, compared to patients with a TBRmax greater than 2.0 (665 days versus 220 days; p < 0.01). Another group used 11C-methionine PET in 22 pediatric patients with diffuse intrinsic pontine glioma (221). 11C-methionine uptake above that of uninvolved brain tissue was noted in 18 of 22 baseline scans (82%) and 15 of 17 initial response scans (88%). However, 11C-methionine PET avidity and intensity did not correlate with survival outcomes.
Cerebrospinal fluid analysis and evoked potentials are rarely indicated for diagnosis of brainstem gliomas in the era of CT and MRI. If cerebrospinal fluid is evaluated, it is usually normal (166). In 15% to 20% of patients, there may be a mild pleocytosis (less than 20 cells) and elevation of cerebrospinal fluid protein. Rarely, malignant cells may be noted on cytologic examination in those patients with leptomeningeal spread of tumor (164). Newer studies are now performing proteomic analyses of cerebrospinal fluid in an effort to better characterize diffuse intrinsic pontine gliomas (201; 169). Numerous specimens from 10 patients were analyzed using various techniques, including proteomics, western blot analysis, and immunohistochemical staining, and selective upregulation of cyclophilin A and dimethylarginase 1 was noted. Auditory evoked potentials have also been used for the diagnosis of brainstem gliomas, although they may be more useful for the monitoring of these patients during neurosurgical procedures (161; 166).
The role of surgical therapy is more limited for patients with brainstem gliomas than for patients with tumors at other sites because of the intrinsic risks of aggressive surgery in this location (166; 142; 155; 49). This is especially true for more diffuse and infiltrative tumors. Despite these restrictions and limitations, surgical intervention is still an important aspect of management for many patients with brainstem neoplasms. The morbidity and mortality associated with surgery of the brainstem have been substantially reduced over the past 15 to 20 years due to improved neuroimaging with MRI, advances in microsurgical instrumentation and technique, use of intraoperative real-time ultrasound and evoked potential monitoring, application of the CO2 laser and ultrasonic aspirator, and improvements in both open and stereotactic approaches to the brainstem (08; 55; 96; 148; 01; 97; 53; 196; 182). In addition, the application of computerized 3-D surgical navigation systems for removal of brainstem tumors has further improved the ability to resect with minimal morbidity (229). The ability of MRI to more accurately define the anatomy and infiltrative extent of brainstem tumors improves the ability of the neurosurgeon to decide if surgery is feasible and, more importantly, if it is appropriate (53; 166). By MRI classification, tumors most amenable to surgical intervention include focal lesions of the midbrain or pons, cystic astrocytomas without cyst wall enhancement, cervicomedullary neoplasms, and the exophytic components of tumors (52; 53; 73; 16; 28; 166; 174; 64; 205; 229; 241). Real-time ultrasound imaging has been reported by many authors as an excellent method during surgery to verify tumor location, depth, and boundaries (55; 196). Variations in echogenicity allow for differentiation of solid tumor from areas of calcification, cyst, and edema. Similarly, the use of evoked potentials (auditory or somatosensory) during surgery allows for real-time monitoring of potential damage to the brainstem and often decreases morbidity (08; 55; 53; 196). Significant improvements in computerized CT- and MRI-guided stereotactic techniques have also contributed to the reduction of surgical morbidity and mortality associated with brainstem tumors. In the majority of reports, surgical mortality was 1% to 3%, whereas morbidity ranged between 3% and 7% (96; 60; 73; 182). A transfrontal or transcerebellar approach to biopsy was used in most of these studies, with diagnostic yields between 90% and 100%.
Although modern neurosurgical techniques allow for acceptable morbidity and mortality when accessing masses within the brainstem of adults, the literature remains unclear about which lesions should be surgically approached and how aggressive that intervention should be. Many authors feel strongly that all brainstem lesions should at least be biopsied because of the high rate (10% to 30%) of nonneoplastic lesions that have been reported in many surgical series (55; 96; 13; 60; 63; 01; 73; 182; 240; 144; 187). Nonneoplastic lesions found at biopsy include chronic inflammation, granulomatous inflammation, gliomatosis, cysts, abscesses, infarction, necrosis, tuberculoma, encephalitis, vascular malformations, and regions of demyelination. Biopsies are strongly recommended to provide definitive histological diagnosis prior to further therapeutic recommendations, especially in adult patients with non-diffuse, contrast-enhancing lesions (187). This approach would prevent patients with nonneoplastic lesions from receiving empiric radiation therapy. A significant counter argument is that biopsies provide small pieces of tissue that may not be diagnostic or that may suggest a lower-grade tumor (166; 05). In tumors as heterogeneous as gliomas, this can be a significant shortcoming of the technique. Other authors do not feel all patients should undergo biopsy or attempted surgical resection, reserving these procedures for carefully selected, prognostically favorable patients. Epstein and colleagues recommend an aggressive approach (ie, radical subtotal excision) for the more "benign" tumors, based on their review of a series of 92 patients classified by MRI and clinical criteria (52; 53). Tumors in this "benign" group usually had low-grade pathology and included focal, nonenhancing cystic and cervicomedullary lesions. Radical subtotal excision of 50% to 60% of focal and cystic tumors and 80% to 90% of cervicomedullary tumors could be achieved with the use of evoked potential monitoring, Cavitron ultrasonic surgical aspirator, laser, and intraoperative ultrasound. Most of these patients were neurologically stabilized or improved following resection and enjoyed prolonged survivals. In contrast, all diffuse tumors and those with contrast-enhancing cyst walls were histologically malignant and did not benefit from surgical intervention, except for cyst drainage in tumors with mass effect. Epstein and colleagues recommend surgery only for the carefully selected "benign" tumors, especially if the intent is to perform a resection. For most diffuse and malignant cystic tumors, the typical clinical presentation and MRI findings will obviate the need for surgical intervention. This view is in agreement with Albright and others, who argue that in patients with a typical history and clinical presentation of a brainstem glioma in which the MRI shows a diffuse lesion, surgical intervention is unnecessary because the histology is virtually always malignant (07; 05; 64; 121). More aggressive, open surgical approaches are advocated by other groups to allow for microsurgical resections within the brainstem (11; 55; 28; 240). However, radical extirpation has only proved possible in low-grade tumors (mainly focal or cervicomedullary) that had pilocytic or grade I/II fibrillary astrocytic histology. Focal tumors of the midbrain are reported by some authors as ideal lesions for an aggressive, open approach for microsurgical resection (172; 171; 123; 229). In 7 adult patients with focal midbrain tumors who underwent aggressive microsurgical resection by Pendl and Vorkapic, all had either stable or improved Karnofsky ratings following surgery (171). The mean survival following surgery was 31.0 months, even though only 2 patients received postoperative radiation therapy. Tumors of the midbrain tectum are also thought to be accessible for resection, although this approach involves more severe mortality (8.3%) and morbidity (33.3%) than a simple biopsy (123). In a series of 35 midbrain gliomas (mean age 26.6 years), Wang and colleagues used a neuronavigation system to perform aggressive resections (229). Total resection was possible in 19 cases. Nine patients with malignant histology (AA or glioblastoma multiforme) received postoperative radiotherapy. Twenty-three of the patients (65.7%) were able to resume independent living. Cystic brainstem gliomas should be considered for cyst drainage if they are not amenable to radical excision, as this may improve localized pressure on brainstem structures (97; 03). In patients with cysts that have recurred following prior aspiration, after external radiation therapy, or that are causing rapid neurologic deterioration, stereotactic intracavitary placement of 32-phosphorus may stabilize further cyst development and result in neurologic improvement (97). Another subgroup of brainstem tumors that are candidates for aggressive, open surgical extirpation is dorsally exophytic lesions (53; 28; 166; 142; 174; 240). These tumors are typically slow growing, histologically benign, and noninfiltrative. The majority of patients with dorsally exophytic tumors can undergo radical subtotal resection and enjoy prolonged survival without further therapy (174). Reports suggest that stereotactic biopsy can be guided by a combination of MRI and PET data (144; 173). In a series of 30 patients with brainstem masses, MRI and PET were considered complementary in determining the proper location for biopsy and obtaining diagnostic tissue. In some cases, PET defined a “hot spot” appropriate for biopsy more clearly than the corresponding MRI. However, in 37% of the cases, MRI and PET were discordant with the pathology and the clinical treatment plan as defined by the final histological diagnosis. The authors concluded that although MRI and PET improve the diagnostic yield when biopsying brainstem mass lesions, neuroimaging alone cannot provide the necessary information to direct treatment in all cases, and histological examination will often be required. An update by the authors, in an additional series of 20 patients, verified the original observations of the utility of PET to improve the targeting of stereotactic biopsy for brainstem gliomas (173). In addition, authors have been using intraoperative MRI for more detailed information regarding surgical approach and extent of resection (120). The use of MRI guidance should be even more helpful in adults with brainstem gliomas because there will be more room to maneuver in the posterior fossa.
Adult patients with dorsal midbrain tumors (ie, tectal) are considered a special, indolent subgroup that should be carefully evaluated before attempting aggressive intervention (241). After a review of 5 patients, Yeh and colleagues concluded that most of these patients can be followed by serial MRI scans and may remain stable without intervention, except for occasional diversionary shunting. Tumor-specific therapy should be reserved for patients with clinical or MRI evidence of growth.
A review of microneurosurgical treatment in a series of 16 consecutive adult patients was not as favorable as some of the results noted above (155). Of the 16 patients, 9 had diffuse pontine tumors, 6 had more craniocaudal-oriented lesions, and 1 had a cystic tumor of the pons. Gross total resection was achieved in only 2 patients, whereas another 9 had subtotal resections. Postoperatively, 11 patients experienced neurologic deterioration, with significant recovery noted in only 4 of the cohort. The authors suggested that the rate of recovery after surgery was much less than what is reported in pediatric brainstem glioma patients treated with surgery. In addition, they questioned the benefit of surgical intervention in adult patients with brainstem glioma.
After reviewing the neurosurgical literature, the current author does not recommend a surgical biopsy or attempted resection in those patients with a classic, diffuse brainstem glioma as defined by MRI. The vast majority of these tumors will have high-grade histology and behave in a malignant fashion. All other brainstem masses should at least undergo an attempted biopsy, due to the high rate of nonneoplastic lesions found in many series. In the hands of an experienced neurosurgeon, carefully selected brainstem tumors (focal midbrain, dorsally exophytic, nonenhancing cystic, cervicomedullary, etc.) can undergo radical subtotal resection.
Once the diagnosis of brainstem glioma has been made, by either surgery or MRI, the mainstay of therapy in most adult patients will be external beam radiation therapy. Early reports from the pre-CT and CT era demonstrated a dose-response effect in brainstem gliomas. Extended survival only occurred in those patients receiving 4500 cGy or more. Additional conventional radiation therapy trials have used focal ports utilizing doses ranging from 4400 to 6500 cGy, with the majority of patients receiving more than 5000 cGy (76; 77; 206; 81; 80; 49; 86). In the study by Shibamoto and colleagues, which included 27 adults (61% with high-grade tumors), 77% of patients with noncontrast-enhancing tumors, 64% of those with diffuse contrast-enhancing tumors, and 50% of tumors with ring enhancement demonstrated a complete or partial response (206). Despite these promising initial results, the overall 5-year survival rate was only 23%. A similar 5-year survival rate of 28.7% was noted by Grigsby and colleagues in a cohort of 27 adult patients (76). In an analysis of 21 adult patients with pontine gliomas, Guiney and colleagues noted clinical or radiological improvement in 77% (81). However, the median survival was only 15 months, with a 3-year survival rate of 33%. In a review of 37 patients with midbrain tumors (16 over 16 years of age) treated with conventional radiation doses of 4500 to 5000 cGy, Franklin noted an overall median survival of 10 months, with a 2-year survival rate of 23% (62). If the patients receiving more than 5000 cGy were analyzed separately, the median survival increased to 14.5 months. A study by Yoshida and colleagues reported the results of radiotherapy in a cohort of 16 adults (median age 49 years) (244). The median radiotherapy dose was 5,600 cGy (range 5,000 to 7,000 cGy); temozolomide was also administered in several patients. The median overall survival for the adult cohort was 39 months, with 1-, 2-, and 3-year overall survival rates of 75%, 68%, and 53%, respectively.
Because of the evidence for a dose-response effect with conventional irradiation and the fact that brainstem tumors are mainly a focal disease, alternative treatment schedules have been studied in an attempt to administer higher total doses without an increase in neurotoxicity (166; 142). The most promising alternative schedule is hyperfractionated radiation therapy, in which a larger number of smaller fractions (eg, 100 to 120 cGy/fraction, twice daily) are administered over an equivalent treatment interval to a larger total dose (eg, 6800 to 7800 cGy). The rationale for hyperfractionation relates to frequency of fractions, dose per fraction, total dose, repair of sublethal radiation damage, and oxygen dependence, which have been reviewed elsewhere (51; 130; 166; 207; 142). Hyperfractionated protocols have demonstrated marginal improvement in overall and progression-free survival in pediatric patients with brainstem gliomas at doses between 7000 and 7200 cGy (166; 142; 86). There have been few reports of hyperfractionated protocols for adult patients with brainstem gliomas. The earliest study was from Edwards and colleagues, who evaluated 19 adults as part of a larger cohort in a phase I-II trial of 100 cGy twice a day to a total dose of 7200 cGy (51). After the completion of radiotherapy, the majority of patients had stable disease as indicated by MRI, whereas less than a third showed reduction in tumor volume. The median survival was 7.6 months, with a median time to tumor progression of 5.5 months. In the study by Linstadt and colleagues, 14 adult patients were treated with hyperfractionated radiation (100 cGy twice a day) to a median dose of 7200 cGy (130). All but 1 patient improved or remained neurologically stable following treatment. The 3-year actuarial survival rate was 59%, with a projected median survival in excess of 5 years. However, there were no glioblastoma multiforme patients in this cohort; most had moderately anaplastic astrocytomas. Shrieve and colleagues analyzed the results of a cohort that included 45 adult patients who received between 6600 and 7800 cGy of hyperfractionated radiotherapy (207). The median survival was 15.8 months, with 1- and 2-year survival rates of 68% and 53%, respectively. Survival was not significantly different for patients who received doses greater than 7200 cGy and those with doses less than or equal to 7200 cGy. When comparing these results with their previous experience in adult brainstem glioma patients using conventional radiotherapy, there was a trend toward improved outcome with hyperfractionation (p = 0.08). Their conclusion was that hyperfractionated radiotherapy was an effective treatment for adults with brainstem gliomas and that 7200 cGy was the preferred dosage. In a review of patients who had received 7800 cGy of hyperfractionated radiotherapy, Prados and colleagues included 45 adults (177). Tumor size, as indicated by CT or MRI, decreased in 30%, was stable in 39.5%, and increased in 29.6%. The median survival of the adult patients was 16.1 months, with a median time to tumor progression of 11.4 months. The median survival was almost identical to that reported by Shrieve and colleagues using 7200 cGy, prompting the authors to conclude that hyperfractionated radiation therapy to 7800 cGy did not improve survival and was associated with greater toxicity. Several studies in adults and children suggest that hyperfractionated protocols do not impart a survival advantage over conventional radiation treatment (24; 64; 203; 137; 86). In a pediatric study by Mandell and colleagues, 132 children with diffuse brainstem tumors were randomized to receive either conventional (180 cGy/day, 5400 cGy total) or hyperfractionated (117 cGy twice per day, 7020 cGy total) irradiation over 6 weeks in combination with cisplatin chemotherapy (137). The median time to progression was 6 months for conventional therapy and 5 months for hyperfractionated treatment. Similarly, median survival was 8.5 months for conventional therapy and 8 months for patients receiving hyperfractionation. In a study from the Mayo Clinic, 40 patients (median age 29.5 years) with brainstem tumors were reviewed for response to irradiation; 9 of the patients received hyperfractionated therapy (203). No survival advantage could be demonstrated in the cohort treated with hyperfractionated radiotherapy.
In an attempt to improve efficacy of radiotherapy, several investigators have begun to use radiosensitizers. Marcus and colleagues added etanidazole, a hypoxic-cell sensitizer, to 66 Gy of hyperfractionated irradiation in a phase I study of 18 children with brainstem glioma (140). The maximum tolerated dose of etanidazole was 42 mg/m2, with cutaneous rash as the dose-limiting toxicity. A similar phase I study by Sanghavi and colleagues used topotecan in combination with conventional radiotherapy for 17 children with brainstem tumors (200). Topotecan is a topoisomerase I inhibitor that is known to convert sublethal radiation-induced DNA damage into lethal double-strand DNA breaks. The maximum tolerated dose of topotecan was 0.40 mg/m2 per day during radiotherapy, with neutropenia as the dose limiting toxicity. A subsequent phase II study of topotecan and concomitant irradiation used the same dosing schedule in a series of 32 patients with malignant diffuse brainstem gliomas (19). In 40% of the cohort, partial responses were noted. However, the responses were short-lived, and the 12-month survival rate was only 25.5%. In addition, the overall median survival was only 8.3 months, similar to other studies of radiation therapy alone. The authors concluded that the addition of topotecan could not be recommended. Another approach has been to add weekly RMP-7 (300 ng/kg) and intravenous carboplatin (35 mg/m2 per day) during the course of irradiation in a phase I trial of newly diagnosed brainstem glioma patients (165). RMP-7 is a bradykinin analog that transiently increases permeability across the blood-brain barrier and may allow higher concentrations of carboplatin to reach tumor cells. The treatment was relatively well tolerated, but no conclusions could be drawn regarding efficacy of the protocol, although the estimated median survival of the cohort was more than 11 months. A similar phase II study using a different dosing regimen of RMP-7 (600 ng/kg) and intravenous carboplatin (target AUC of 3.5 mg per min/ml per day) was applied to 12 patients with brainstem and other high-grade gliomas (231). Although the regimen was relatively well tolerated except for hematological toxicity, none of the brainstem glioma patients responded.
Although brachytherapy has demonstrated improved survival for patients with supratentorial gliomas, few authors have applied this technique to brainstem tumors because of the inherent dangers of radiation exposure in this location. Mundinger and colleagues have used iodine-125 and iridium-192 as temporary or permanent radiation sources for 55 patients (mean age 22.5 years) with nonresectable low-grade brainstem gliomas, most of which were focal lesions of the midbrain (154). The majority of patients (61%) had pilocytic astrocytomas, whereas all others had low-grade protoplasmic or fibrillary astrocytomas. The 5-year survival rates were 54.8% for the iodine-125 group and 26.9% for the iridium-192 group; these rates were marginally significantly different (p = 0.03). Mortality for the procedure was 2.4%, with a morbidity of 6.2%.
Radiosurgery is a technique that is able to deliver photon radiation to well-circumscribed volumes of tumor deep within the brain. Several authors have applied it to patients with brainstem gliomas (133; 112; 204; 66; 49; 242; 41). In their series of 7 patients with midbrain tectal low-grade gliomas who ranged in age from 6 to 40 years (mean 18 years), Kihlstrom and colleagues administered doses of 1400 to 3500 cGy in 1 fraction using a single isocenter (112). In 5 of the 7 cases, the tumor shrank progressively after treatment. All 7 patients remain alive, with mean follow-up time of 6 years. Patients treated with the higher doses (eg, 1800 to 3500 cGy) developed significant complications, including radionecrosis and progression of neurologic deficits. The authors concluded that radiosurgery was useful adjunctive therapy in this group of tumors when using a recommended dose of 1200 to 1400 cGy. Loeffler and colleagues, in their series of 41 malignant glioma patients treated with radiosurgery, included 1 adult with a brainstem glioblastoma multiforme (133). That patient received 1000 cGy to the 80% isodose line following external beam radiotherapy. Treatment has resulted in a partial response, with clinical improvement and cessation of steroids after 6 months of follow-up. In a series of 41 patients with brainstem glioma (26 adults), Schulz-Ertner and colleagues used fractionated stereotactic conformal radiosurgery with a linear accelerator (204). The field was shaped with a multileaf collimator, for a median total dose of 5400 cGy to the 90% isodose line. Clinical improvement was noted in 19 of 41 patients (46.3%), whereas objective radiological responses occurred in 12 of 41 patients (29.3%). Median time to progression was 23 months, with a median overall survival of 40 months. The treatment was tolerated well and the authors concluded that this approach was superior to conventional irradiation with parallel-opposed ports due to the use of a smaller, more precise treatment volume. Fuchs and colleagues used gamma knife radiosurgery with a median dose of 12 Gy for a series of 21 patients with brainstem gliomas that were mostly adult (median age 23 years) (66). Ten patients responded with stable disease, whereas 3 had objective responses. Clinical improvement was noted in 5 patients. The median overall survival was 20.7 months. One report describes the use of Gamma Knife as the primary method of treatment in 20 patients with focal brainstem lesions (242). There were 4 complete responses and 12 partial responses/minor responses during a mean follow-up period of 78.0 months. The remaining patients were stable or clinically improved.
A report by Combs and colleagues described the results of using fractionated stereotactic radiosurgery in a cohort of 85 patients with brainstem gliomas (41). The median age at primary diagnosis was 26 years of age, with only 31 patients under 18 years of age. A median dose of 54 cGy was administered in 1.8 Gy fractions to a median target volume of 101 mL. The median progression-free survival was 52 months, with progression-free survival rates at 12 and 24 months of 70% and 63%, respectively. Radiosurgery was well tolerated and did not appear to have a dose-response relationship in this group of patients. Susheela and colleagues describe their experience with reirradiation of progressive brainstem glioma patients using a hypofractionated stereotactic robotic radiotherapy approach (216). Five progressive adult brainstem glioma patients were reirradiated with doses ranging from 16-25 Gy, delivered in 2 to 5 fractions. In 4 of the 5 patients, there was prolongation of survival ranging from 3 to 14 months, with stable neurologic function. In the fifth patient, there was prolonged survival out to 36 months. However, in more recent months there have been signs of radiotherapy-induced brainstem toxicity.
Patients who are suspected of having leptomeningeal metastasis from a brainstem glioma should be evaluated on an individual basis and, if clinically indicated, undergo an evaluation of CSF (including cytology and tumor markers such as beta-glucuronidase and beta2-microglobulin), MRI of the spine with gadolinium, and myelography (164; 166). In patients without evidence of infection, abnormal levels of tumor markers can improve diagnostic sensitivity. Patients with proven leptomeningeal seeding should then be considered for craniospinal axis irradiation and intrathecal chemotherapy. Typical doses for craniospinal axis irradiation are 3000 to 3500 cGy. In patients with clinical and imaging evidence for focal bulky involvement by leptomeningeal tumor, radiation can be administered to a localized field with generous margins, using doses of 3500 to 4500 cGy.
Over the past 2 decades, chemotherapy has had little impact on the overall survival of patients with brainstem gliomas (166; 142; 65; 49; 86). The vast majority of data regarding chemotherapy are from the pediatric literature, as summarized by Maria and colleagues (142). A compilation of 18 studies using various chemotherapy regimens for brainstem tumors in children encompasses 261 patients. Of this cohort, 48 patients (18.4%) responded, including 21 with partial responses and 27 with stable disease; there were no complete responses. The most effective single agents were carboplatin and cyclophosphamide. Unfortunately, all of the responses were brief, as tumors rapidly progressed through chemotherapy. A phase III trial of irradiation with or without adjuvant lomustine, vincristine, and prednisone was reported by Jenkin and colleagues (104). The overall 5-year survival rate was 20% for both groups, suggesting that chemotherapy did not significantly affect outcome. In the study by Levin and colleagues, 28 patients (including 2 adults) were treated with 5-fluorouracil and lomustine before irradiation, followed by hydroxyurea and misonidazole during radiotherapy (129). The median survival was 11 months, which, when compared with prior trials, was no better than radiation therapy alone. Similarly, a Children’s Cancer Group study combining hyperfractionated radiotherapy (7200 cGy) and intravenous recombinant beta-interferon did not demonstrate improved efficacy over conventional radiotherapy alone (median time to progression 5 months; median survival 9 months) (167). Several studies have attempted high-dose chemotherapy with autologous bone marrow transplantation in pediatric brainstem glioma patients (25; 26; 50). Bouffet and colleagues used high-dose BCNU (800 mg/m2) on 8 patients with brainstem glioma (25). One patient briefly stabilized for a few months whereas the other 7 patients rapidly progressed. In a similar study of 24 children with diffuse pontine gliomas, Bouffet and colleagues used high-dose busulfan (150 mg/m2) and thiotepa (300 mg/m2) followed by autologous bone marrow transplantation (26). The median time to clinical progression was 5 months, with an overall median survival of 10 months. Dunkel and colleagues used high-dose thiotepa and etoposide, either alone or in combination with BCNU or carboplatin, in 16 patients with diffuse pontine tumors (50). The median survival was 4.7 months for patients with recurrent disease and 11.4 months for patients with newly diagnosed tumors. High-dose therapy did not improve survival over conventional treatment and was associated with considerable toxicity. A Children’s Cancer Group study used intravenous idarubicin (5 mg/m2 per day for 3 days, every 21 days) for patients with relapsed brain tumors, including 13 with brainstem gliomas (12). None of the 12 brainstem glioma patients able to be evaluated responded to treatment. In an attempt to increase dose intensity for newly diagnosed malignant glioma patients, Jakacki and colleagues used time-compressed, dose-intensive lomustine, procarbazine, and vincristine in combination with peripheral blood stem-cell support and concurrent irradiation (101). Of their cohort of 6 patients with diffuse pontine brainstem gliomas, 2 had partial responses and 1 had a minor response. The median time to radiographic progression was 5 months, with a median overall survival of 11 months. In a similar study of 22 assessable patients with newly diagnosed brainstem gliomas, Broniscer and colleagues used a combination of irradiation and concurrent tamoxifen (200 mg/m2 per day) (32). There were 8 partial responses, 3 minor responses, and 8 patients with stable disease. The overall median survival was 10.3 months. Allen and colleagues performed a phase I/II trial of intravenous carboplatin and hyperfractionated irradiation (100 cGy twice a day; 7200 cGy total dose) in 34 children with brainstem gliomas (09). The maximum tolerated dose of carboplatin was 110 mg/m2 over 7 weeks of radiation. The median progression-free survival was 8 months and the overall survival was 12 months. A phase II study by the Children’s Cancer Group compared preirradiation carboplatin, etoposide, and vincristine versus cisplatin, cyclophosphamide, etoposide, and vincristine (106). Neither regimen was able to improve response rate, time to progression, or overall survival relative to prior studies of brainstem glioma patients treated with radiotherapy with or without chemotherapy. In a phase II study by Burzynski and colleagues, intravenous bolus injections of antineoplaston A10 and AS2-1 were administered to a series of 12 patients with recurrent brainstem gliomas, including 1 adult and 4 teenagers (36). Five patients had brief objective responses, for a 2-year survival rate of 33.3%. Ronghe and colleagues reported their experience using vincristine and carboplatin for pediatric patients with unresectable and/or recurrent low-grade astrocytomas of the brainstem (193). Sixteen patients received treatment, and all were still alive at the time of the report, with 9 objective responses on neuroimaging (1 complete response, 8 partial responses/minor responses), and another 4 that remained stable. Overall, 68% of the cohort remained progression-free, with a median follow-up of 57 months.
Studies by the Pediatric Brain Tumor Consortium have investigated the potential efficacy of molecular-based chemotherapy drugs in children aged 3 to 21 (107). Pollack and colleagues reported the results of a phase II study of gefitinib, an oral inhibitor of EGFR, when used during and after radiation therapy (176). Forty-three eligible patients received gefitinib (250 mg/m2/day) during irradiation, and afterwards as adjuvant therapy. The 12- and 24-month progression-free survival rates were 20.9% and 9.3%, respectively, whereas overall 12- and 24-month survival rates were 56.4% and 19.6%, respectively. The authors concluded that this regimen had activity and might be even more effective in patients positive for EGFR on molecular analysis. In the study by Hass-Kogan and colleagues, newly diagnosed patients with diffuse pontine gliomas were treated with tipifarnib, an oral farnesyltransferase inhibitor, during and after radiotherapy (88). Forty eligible patients received tipifarnib during radiotherapy (125 mg/m2/dose bid) and as adjuvant therapy (200 mg/m2 bid, 21 days on and 7 days off). The median progression-free survival was 6.8 months, with a median overall survival of only 8.3 months. The authors concluded that tipifarnib, used during and after radiotherapy, did not offer any advantage over historical treatment approaches.
Few chemotherapy studies of brainstem gliomas have included more than a few adult patients. However, the experience with chemotherapy in adults has been similar to that of children, with few durable responders (121; 203). In a review of 21 patients treated with 5-fluorouracil, lomustine, hydroxyurea, and 6-mercaptopurine at progression after radiation therapy, Rodriguez and colleagues included 4 adults (191). The ages ranged from 16 to 25 years, with a mean of 19.75 years. The overall mean survival of the adult cohort was 11.9 months, with a mean time to progression of 5.1 months. Chamberlain has used chronic oral etoposide (50 mg/m2 per day for 21 days every 5 weeks) to treat 12 patients, including 4 adults (age range 18 to 49 years), with recurrent brainstem gliomas (37). The overall results demonstrated 6 radiographic responses with a median duration of response of 8 months. Of the adult patients, 2 had partial responses of 4 and 12 months' duration, whereas the remaining 2 had progressive disease. In a group of 3 adult patients (age range 18 to 59 years) with pontine gliomas given chemotherapy after irradiation, Fujiwara and colleagues used monthly intraarterial ACNU (80 mg/m2), cisplatin (40 mg/m2), or carboplatin (300 mg/m2) in addition to oral etoposide (25 mg/day) and intravenous beta-interferon (67). One patient with a grade II astrocytoma had a complete response after 8 cycles of intraarterial carboplatin and remained in remission 24+ months after diagnosis. Another patient with an anaplastic astrocytoma had a partial response to ACNU that lasted through 4 cycles of treatment, followed by progression and expiration 14 months after diagnosis. The last patient had a grade II astrocytoma that did not initially respond to ACNU, but did demonstrate a partial response to intraarterial cisplatin that lasted for 3 cycles (survival 15 months). The authors concluded that this regimen was well tolerated and effective in brainstem glioma patients. In 2 studies of adult brainstem glioma patients reported from Memorial Sloan-Kettering Cancer Center and the Mayo Clinic, chemotherapy was unable to significantly impact survival (121; 203). Trofosfamide is an alkylating agent with a similar structure to cyclophosphamide but with improved lipid solubility and excellent enteral absorption. Wolff and colleagues were unable to demonstrate any survival benefit in a trial of trofosfamide (100 mg/m2 per day) and etoposide (40 mg/m2 per day) when added to irradiation in a group of pediatric patients with glioblastoma multiforme, including 8 with tumors of the brainstem (235). These results were corroborated by the same group in another study with an enlarged cohort of 20 patients with pontine glioma (236).
A multi-institutional study evaluated the use of temozolomide (200 mg/m2 x 5 days every 28 days for 6 cycles) after the completion of irradiation in newly diagnosed patients with brainstem glioma (30). Thirty-three patients were treated, including several young adults. The median time to progression was 8.8 months, with a median overall survival and 1-year survival rate of 12 months and 48%, respectively. The authors concluded that temozolomide was not very active in this group of patients and did not improve their poor prognosis. An Italian study in adult patients with brainstem glioma seems to corroborate this pediatric data. Salmaggi and colleagues treated 18 adults with chemoradiation using temozolomide (75 mg/m2/day), followed by up to 9 cycles of adjuvant temozolomide (200 mg/m2/day x 5 days). The addition of temozolomide to irradiation did not seem to add any survival benefit and did not result in any complete or partial responses (198). The lack of improved treatment response when temozolomide is added to radiotherapy has been verified in a report of a group of pediatric brainstem glioma patients (38). However, occasional case reports still suggest that temozolomide can be active in a small subpopulation of brainstem glioma patients. For example, a young adult with a brainstem glioblastoma multiforme that had an objective response on MRI survived for more than 3 years (248). In addition, a European study of 15 adults with low-grade diffuse pontine brainstem gliomas suggests some activity from temozolomide (189). All of the patients had low-grade histology (grade II oligodendroglioma - 2, grade II oligoastrocytoma - 2, grade II astrocytoma -1) or were considered low grade by MRI criteria and had failed standard irradiation. Each patient received temozolomide (150 to 200 mg/m2/day x 5 days) every 28 days. Clinical improvement was noted in 9 patients (60%), with radiological responses noted in 6 cases, including 4 partial responses and 2 minor responses. The estimated median progression-free survival after temozolomide was 9.5 months, with a median overall survival of 14.4 months. Grade 3 thrombocytopenia was present in 26 of the cases. Hall and colleagues report an aggressive approach to chemotherapy of diffuse pontine glioma, using intraarterial carboplatin or methotrexate in combination with blood-brain barrier disruption (BBBD) and intravenous cyclophosphamide and etoposide (84). Eight patients (1 adult, 2 teenagers) received the regimen, resulting in 2 partial responses and 5 stable diseases, with a median time to tumor progression of 15 months. The overall median survival of the cohort was 27 months.
Bevacizumab is a chemotherapy drug approved by the U.S. Food and Drug Administration for treatment of recurrent glioblastoma (158). It is a humanized monoclonal antibody targeted against vascular endothelial growth factor and has potent antiangiogenic and antiedema activity. Several case reports suggest that bevacizumab may have activity in some adult patients with brainstem gliomas (183; 224). In the first report by Torcuator and colleagues, the authors described a 43-year-old woman with a diffuse pontine tumor who initially received chemoradiation with temozolomide (224). She progressed after only 1 cycle of adjuvant temozolomide and was then started on combination therapy with bevacizumab and irinotecan every 2 weeks. After 8 months of treatment, the patient experienced improvement in neurologic function, and the MRI scan showed some reduction in FLAIR abnormality and enhancement. Irinotecan was stopped because of fatigue and neutropenia; bevacizumab was continued for another 4 months. The patient has remained stable off treatment. Raza and Donach described an adult patient with a newly diagnosed brainstem glioma who received radiotherapy in combination with temozolomide and bevacizumab, followed by adjuvant treatment with both drugs (183). This patient also had improvement of FLAIR and enhancement on MRI scan, and now remains stable off treatment after 30 months. Another case report describes the use of intraarterial bevacizumab, in combination with mannitol-induced disruption of the blood-brain barrier, in a 43-year-old male with a recurrent malignant brainstem glioma (190). The procedure was well tolerated, with postprocedure imaging showing a reduction in enhancement within the tumor. There were no procedural or vascular complications from the treatment. In a report from the Pediatric Brain Tumor Consortium of 17 patients with recurrent diffuse pontine glioma, the combination of bevacizumab and irinotecan did not appear to have any significant activity in terms of clinical stabilization or by neuroimaging criteria (82). Several case reports have also used bevacizumab in combination with other chemotherapy drugs for treatment of brainstem glioma. In the report from Aguilera and colleagues, 2 patients with diffuse pontine brainstem glioma received irradiation, followed by temozolomide (200 mg/m2 per day x 5 days per month) and bevacizumab (10 mg/kg per dose every 14 days) (02). Follow-up MRI scans demonstrate a greater than 65% reduction in size of the tumors in each patient. The ongoing progression-free survival is now greater than 37 and 47 months from initial diagnosis. Both patients were able to discontinue steroids by 10 weeks after radiotherapy. Blesa and colleagues describe a patient with a relapsed high-grade glioma involving the brainstem that had an excellent response to bevacizumab and cetuximab (monoclonal antibody directed against EGFR) (22). Following treatment with this combination in the third-line setting, the patient has had a complete radiological response that has been durable for 20 months.
Based on the pathology data documenting the presence of EGF receptors in diffuse pontine gliomas, several authors have begun to treat patients with nimotuzumab, a humanized IgG1 monoclonal antibody designed to target EGFR (145; 146). Several pediatric patients treated with the drug have shown objective MRI responses and extended survival, including an 8-year-old boy who had tumor shrinkage that was maintained for more than 8 months (146).
Animal work by Sandberg and colleagues suggests that convection-enhanced delivery of chemotherapy may be applicable to brainstem gliomas (199; 147). In studies using rats, it was determined that 1-time convection-enhanced chemotherapy infusions could be safely applied to the brainstem. Convection-enhanced delivery bypasses the blood-brain barrier and can achieve high local concentrations of infusate. Further studies will evaluate the tolerability of multiple infusions and the use of various forms of chemotherapy. Another series of studies in rats has tested the safety and tolerability of interstitial infusion of IL13-PE38QQR into the brainstem (212). IL13-PE38QQR is a recombinant immunotoxin comprised of a modified Pseudomonas exotoxin moiety conjugated to the human derived T-cell cytokine, IL-13. Infusion of IL13-PE38QQR into the pons at a concentration of 10 µg/ml for up to 2 weeks was well tolerated without clinical or histological toxicity. Similar studies in primates have evaluated the safety and toxicity of chronic carboplatin infusions directly into the pons (214). Overall, infusion of carboplatin up to 0.26 mg/ml for 1 month was well tolerated without significant clinical or radiographic toxicity. Neurotoxicity was noted for infusion doses of 2.6 mg/ml for 1 month. Further studies from the same authors in a monkey model have infused carboplatin into the pons at 0.42 µl/h to deliver 0.025, 0.075, 0.25, and 0.75 mg/kg by day 30 (215). Lethargy and ataxia were noted after 2 weeks in animals given 0.25 mg/kg and after 4 weeks in animals given 0.075 mg/kg. Lower doses were tolerated well without systemic or neurotoxicity. Studies of convection-enhanced delivery using a rat model have verified the safety of infusion using 1, 2, or 3 catheters. However, when the catheters were used to infuse carboplatin, drug-related neurotoxicity was noted. Mild neurologic deficits that developed in the animals after cannula implantation resolved in the placebo group but were more permanent in the carboplatin treated animals. The authors stressed the development of less neurotoxic chemotherapy drugs to be used for this kind of regional chemotherapy (220). More recent reviews of the potential for convection-enhanced delivery in patients with diffuse-intrinsic pontine gliomas remain positive and suggest significant potential (253). There are several active clinical trials in patients with diffuse-intrinsic pontine gliomas using targeted macromolecules, including antibodies and immunotoxins. In a phase I trial, Heiss and coworkers infused an IL-13 pseudomonas exotoxin into a cohort of 5 patients with diffuse intrinsic pontine glioma (90). There was a temporary arrest of tumor growth noted in 2 patients. However, by 12 weeks all patients had tumor progression. There were no long-term side effects or toxicity from the convection enhanced delivery infusions. A report by Souweidane and colleagues describes a phase I trial of a radioimmunotherapy agent targeting the glioma-associated B7-H3 antigen in 28 patients with diffuse intrinsic pontine glioma (211). The maximum tolerated dose has not yet been reached. Convection enhanced delivery has been fairly well tolerated; the trial remains ongoing with an expanded cohort.
New molecular-based therapeutic approaches are also under development for brainstem glioma (107; 147). For example, Barton and colleagues used PD-0332991, a CDK4/6 inhibitor, to treat brainstem glioma in vitro and in in vivo animal models (18). The drug was able to induce cell cycle arrest in brainstem glioma tumor cells from PDGF-B;Ink4a-ARF deficient mice. A 7-day course of PD-0332991 was able to significantly prolong survival by 12% in the PDGF-B;Ink4a-ARF deficient mouse model. In addition, a single dose of 10 Gy irradiation followed by 7 days of PD-0332991 was able to increase survival in the mice by 19% in comparison to radiotherapy alone. PD-0332991 was not active in brainstem glioma from a PDGF-B;p53 deficient mouse model. Another approach by Truffaux and coworkers used dasatinib, a multikinase inhibitor, against several cell lines derived from biopsied diffuse pontine gliomas (225). A panel of 12 new DIPG cell lines were established and characterized by cellular and molecular analysis, then treated with dasatinib. Dasatinib treatment at 1 to 10 μM resulted in significant reductions in tumor cell proliferation and invasion. In addition, the activity of downstream signal transduction effectors was strongly reduced. Treatment using dasatinib in combination with cabozantinib (c-MET inhibitor) resulted in a mildly synergistic enhancement in antitumor effects. A similar study by Miyahara and colleagues used a dual mTOR kinase inhibitor against diffuse-intrinsic pontine glioma cell lines (151). Treatment with MLN0128 induced significant reductions in cell growth and proliferation, along with increased tumor cell apoptosis. Another preclinical study evaluated the use of panobinostat--a histone deactelylate inhibitor--in vitro and in animal models (93). Panobinostat was able to potently inhibit cell proliferation, viability, and clonogenicity as well as induce apoptosis in human and murine diffuse-intrinsic pontine glioma cell lines. In addition, in a brainstem glioma mouse model (driven by PDGF-B, p53 mutation, and ectopic H3.3-K27M or H3.3-WT expression) and an H3.3-K27M orthotopic diffuse-intrinsic pontine glioma xenograft model, the drug was able to reach tumor tissue in adequate concentrations and was able to reduce tumor cell proliferation and increase levels of H3 acetylation. However, due to systemic toxicity and necessary dosage reductions, panobinostat did not prolong overall survival in comparison to control animals.
The current author has used chemotherapy to treat 3 adult patients with brainstem glioma who ranged in age from 33 to 63 years (mean 47.3 years). The first patient was a 63-year-old man with a right midbrain anaplastic astrocytoma that remained stable for 7 months during monthly intravenous carmustine (200 mg/m2).
The tumor was unchanged when the patient expired from aspiration pneumonia. The second patient is a 47-year-old woman with a pilocytic astrocytoma of the left middle cerebellar peduncle that had progressed before and after radiation therapy. She has been treated with oral lomustine (110 mg/m2) and intravenous vincristine (1.4 mg/m2) every 6 weeks and remains stable during the first 9 months of therapy. The third patient is a 32-year-old woman with a large brainstem anaplastic astrocytoma centered in the pons.
She has been treated with 6 monthly cycles of intraarterial carboplatin (200 mg/m2 for 2 days) and intravenous etoposide (100 mg/m2 for 2 days), resulting in a partial response (defined as a 50% to 99% reduction in tumor volume on CT or MRI).
The management of the general aspects of care of adult patients with brainstem gliomas is similar to that of patients with other brain tumors (157). In those rare patients with seizures, adequate control can usually be achieved with first-line agents like phenytoin or carbamazepine. Monotherapy should always be attempted with both agents before considering a second-line agent such as valproic acid, gabapentin, or phenobarbital. Because the majority of brainstem glioma patients have some degree of elevated intracranial or localized tumor-related pressure, dexamethasone should be prescribed in the lowest dose that will improve symptoms (eg, 1 to 4 mg 4 times daily) (166; 157). Following stabilization of pressure-related symptoms and initiation of more definitive therapy, dexamethasone should be carefully tapered to minimize potential complications (157). Dysphagia is a relatively common problem in brainstem glioma patients because of tumor-related damage to medullary swallowing centers (157). In all patients with symptoms of dysphagia, regardless of the apparent severity of the complaints, a thorough swallowing evaluation should be performed. Patients with incipient or well-established hydrocephalus should be evaluated and, if appropriate, treated with a CSF diversionary shunt procedure (166; 142).
Researchers are beginning to investigate the use of immune approaches in patients with brainstem gliomas. Benitez-Ribas and colleagues performed a phase 1b trial of autologous dendritic cell vaccination in a series of 9 patients with newly diagnosed diffuse intrinsic pontine glioma (17). Five doses of the vaccine were administered intradermally during the induction phase and could be continued every 3 months in the maintenance phase. Vaccine administration was safe and well tolerated, and resulted in a positive immune response in the majority of patients. Vaccine efficacy will need to be tested in future clinical trials.
Pregnancy does not affect the clinical behavior of brainstem gliomas.
One report described a case of a female patient who was pregnant and was diagnosed with a brainstem glioma during the pregnancy (195). The report also reviewed the literature, and 12 articles were noted describing 17 pregnancies in 16 women; the median gestational age at presentation was 23 weeks. All but one patient presented with neurologic deficits; MRI was the diagnostic test of choice. There were no reported sequelae of maternal oncological management on neonatal wellbeing. However, the maternal mortality rate was high, 50% (8/16), both during (N = 5) and within 4 weeks of pregnancy (N = 3). Pregnancy losses included 1 pregnancy termination and 4 miscarriages (associated with maternal mortality). The authors concluded that appropriate management, including surgical resection and radiotherapy, should not be delayed on account of the pregnancy.
Several anesthetic concerns specific to brainstem gliomas are not typically encountered with other brain tumors. In rare patients, especially adults, tumors may involve regions within the medulla or pons that affect respiration and blood pressure control (192; 98; 222; 226). Apnea, hypoventilation, and hyperventilation may occur when a brainstem tumor damages the lower brainstem pathways and nuclear regions responsible for respiratory control (192; 117; 222; 226). The alterations in PaO2, PaCO2, and pH caused by these respiratory conditions may affect the approach to anesthesia. Abnormal blood pressure control, especially orthostatic hypotension and secondary cardiac syncope, can also develop in these patients due to disturbances of central autonomic control and the afferent arc of the baroreceptor reflex (98). In those patients particularly at risk, cardiac pacing should be considered.
In addition, concerns regarding the presence or absence of elevated intracranial pressure during the induction, maintenance, and emergence from anesthesia are common to surgical therapy of any brain tumor (125). In patients with brainstem gliomas who have elevated intracranial pressure, care should be taken with premedications to avoid agents that produce excessive sedation and ventilatory depression, as these could exacerbate intracranial pressure. Hypotonic fluids should also be avoided whenever possible. During the induction and maintenance of anesthesia, agents should be chosen that minimize hypertension, cerebral vasodilation, chest wall rigidity, and hypercapnia (125).
The author would like to thank Dr. Carl Boesl for his efforts in preparing the neuropathological slides and Mr. David Carpenter for expert editorial assistance.
Herbert B Newton MD
Dr. Newton, Director of the Neuro-Oncology Center at Advent Health Cancer Institute Orlando, has no relevant financial relationships to disclose.See Profile
Rimas V Lukas MD
Dr. Lukas of Northwestern University Feinberg School of Medicine received honorariums from Novocure for speaking engagements, honorariums from Novocure for advisory board membership, and research support from BMS.See Profile
Nearly 3,000 illustrations, including video clips of neurologic disorders.
Every article is reviewed by our esteemed Editorial Board for accuracy and currency.
Full spectrum of neurology in 1,200 comprehensive articles.
Stroke & Vascular Disorders
In this article, the author reviews current knowledge about intracranial aneurysms due to infectious and neoplastic causes. Direct mural injury or invasion
Jul. 21, 2021
This article reviews the diverse systemic manifestations of von Hippel-Lindau disease and discusses the historical context of the disease as well as
Jul. 09, 2021
Carcinomas of the lung or breast are the most common sources of brachial plexus metastases, whereas lumbosacral plexus metastases most often arise from primary pelvic tumors or lymphoma. Local and radiating pain is the most common presenting symptom of nerve plexus metastasis, eventually followed by motor and sensory deficits. MRI is fairly sensitive and specific in diagnosing nerve plexus metastases.
Jul. 06, 2021
Jun. 22, 2021
Paraneoplastic limbic encephalitis is a subset of a larger group of autoimmune encephalitides characterized by the predominant involvement of the limbic system. Patients present with varied combinations of short-term memory loss, seizures, and psychiatric disturbance. Small cell lung carcinoma, thymoma, testicular germ cell tumors, and Hodgkin disease are the tumors most often associated with paraneoplastic limbic encephalitis.
Jun. 01, 2021
Hemangioblastoma is a relatively uncommon tumor that most often arises in the cerebellum, brainstem, or spinal cord. Hemangioblastomas usually occur
May. 25, 2021
Headache & Pain
Apr. 18, 2021
Dysembryoplastic neuroepithelial tumors are rare, indolent, low-grade tumors found in children and young adults. Most commonly affecting the temporal
Mar. 22, 2021