Brain stem and cervicomedullary tumours are typical of paediatric age, 80% of them occurring in patients under 18 years of age, and comprising 10–15% of all childhood and adolescent brain tumours, as well as 20–25% of infratentorial locations.

They are characteristically pontine tumours (60% of the cases), but they commonly extend to involve the medulla, midbrain and cerebellum. Although most brain stem tumours are low grade gliomas, their prognosis is extremely severe (no more than 20% of patients are alive 3 years after diagnosis and the 5-year survival rate is 5%) because both the typical infiltrating nature and the neuro-anatomical location usually make them surgically unresectable. Surgery is generally limited to biopsy, partial decompression, or excission of the exophytic components, because of the extremely severe functional sequelae of even minor resections. Thus, the mainstay of therapy has been based on irradiation alone or combined with chemotherapy, doses of 5000–5500 cGy being usually adequate for tumour shrinkage or remission, even if recurrence is common after 10–15 months.

At present, it is very important to establish reliable, homogeneous, objective, and reproducible diagnostic criteria for the identification of patient subsets with predictable histology, prognosis and possible therapeutic management, in some cases histology, site and relationship of tumour enabling total or subtotal resection with a lower operative risk. Since its introduction, magnetic resonance imaging (MRI) has appeared the procedure of choice for the neuroradiologic study of the brain stem and brain stem tumours, enabling a more precise definition of their margins, a correct assessment of intrinsic and exophytic components, as well as a satisfactory characterization of pathologic tissue. MRI studies should include good quality T1-, PD- and T2-weighted images and T1-weighted images after gadolinium i.v., T2-weighted sagittal images being required for complete evaluation of tumour extent. Computed tomography is still superior in the identification of calcifications and acute intratumoral haemorrhage; it is rapidly performed, thus representing the first choice procedure in emergency, i.e. the diagnosis of hydrocephalus. Epstein has proposed the most widely accepted classification system of brain stem tumours, essentially based on neuroradiological findings, surgical and stereotactic biopsy and histology generally resulting in understaging. This classification system separates intrinsic (diffuse, focal, cervico-medullary), exophytic (anterolateral into cerebellopontine angle, posterolateral into brachium pontis, posterior into fourth ventricle) and cerebrospinal fluid seeding (positive cytology or myelography) tumors.

More recently, Barkovich – based on a multicentric study of some of the most important paediatric neurosurgery and neuro-oncology centres of the United States – has clearly defined the neuroradiological parameters which must be considered for an objective and reproducible assessment of brain stem gliomas, in order to identify patient subsets characterized by predictable histology, prognosis and possible therapeutic management. We agree with him, emphasizing that the evaluation of brain stem tumours must include a careful interpretation of all MRI findings (tumour site and origin; dimensions/degree of brain stem enlargement; tumour caudo-cranial and transverse extension; exophytic components; tumour characteristics as defined by MRI signal intensity; cysts, haemorrhage, necrosis, calcifications; ventricular dimensions and hydrocephalus; leptomeninengeal seeding) that can help in the definition of the following tumour subsets: diffuse pontine tumours, medullary tumours, cervicomedullary tumours, focal brain stem tumours.

With regard to the neuroradiological follow-up, in patients undergoing surgery (for biopsy decompression or less frequently for radicality), the role of the neuroradiologist is similar to that in other fields of neurosurgery, and concerns the quantitation of the extent of the resection and the identification of possible parenchymal injuries or postoperative haemorrhage, always keeping in mind the negative effects of postoperative reactive phenomena and the blood-brain disruption 24–48 hours to 30–40 days after surgery. However, the major contribution of the neuroradiologist is the objective evaluation of irradiation effects and recently of combined irradiation and chemotherapy. In our experience, the efficacy of irradiation is well evaluated by MRI only 3 or 6 months after the end of treatment, even if a clinical improvement is possible after 30–60 days. However, tumour shrinkage is rarely drastic, and its disappearance, as well as that of signal alterations, is exceptional. Qualitative modifications occurring within the tumour are more difficult to interpret, because of the appearance of cystic or pseudocystic areas, markedly increased enhancement, and small areas of haemorrhage may be related both to irradiation-induced regressive modifications and disease progression.

In conclusion, MRI represents the gold standard in the evaluation of brain stem and cervicomedullary tumors, always enabling a precise definition of tumour site and extent, and in most cases the diagnosis of nature, thus allowing the identification of patients who can undergo radical microsurgery. MRI follow-up controls the extent of resection and the effect of combined irradiation and chemotherapy; disease progression is evidenced and spinal seeding can be diagnosed.

However, the differentiation between tumour recurrence and irradiation-induced injury may be difficult if only based on morphological data. These limitations of MRI will probably be reduced by the advances in ultra-fast MRI technology and 18F-fluoro-deoxy-glucose positron emission tomography which supplies in vivo metabolic and functional information.