- Email: [email protected]
Tumours of the central nervous system
NEUROSURGERY
Tumours of the central nervous system
tumours. There are some similarities in the tumour types that affect adults and children, but the differences have also taught us much about the biology and aetiology of these tumours. For one paediatric tumour, medulloblastoma, progress in our understanding of the tumour’s molecular phenotype has pioneered an approach to personalized medicine where the molecular phenotype impacts on treatment and prognosis. A better understanding of the biology of other brain tumours may also help us to improve stratification of affected patients for therapy, which in turn could inform the allocation of different treatments with improved outcomes. We are not there yet, but insights into tumour biology have already driven an updated system of tumour classification that is improving on traditional histological diagnosis. Tumours in the brain may be either primary or secondary tumours. Primary brain tumours are those that develop from cells in the brain. Secondary, or metastatic, tumours are those that have spread from elsewhere in the body. The most common tumours that metastasize to the brain are breast, renal, melanoma and colorectal tumours. There are 17 broad categories of primary CNS tumour according to the WHO 2007 classification, each with multiple subcategories. However, only a handful of the possible tumour types account for most of the tumours that are seen in clinical and surgical practice. The most common primary brain tumours are glioma and meningioma. In 2016 the World Health Organization (WHO) classification of tumours of the CNS was updated according to a new molecular nomenclature.1 This superseded the 2007 classification that was based on histological tumour features. The histology descriptors are based on similarities between tumour cells and the appearance of normal brain cells.1,2 (Box 1). For example, a glioma whose constituent cells look like astrocytes will be an astrocytoma, but if the cells look like oligodendrocytes then is it an oligodendroglioma. The problem arises though that some
Paul M Brennan
Abstract Surgery plays a key role in the management of brain tumours. The prognosis for a brain tumour diagnosis remains poor, but without the input of the surgical team, the outcomes would be bleaker. Innovations in understanding of the molecular biology of brain tumours may identify new targets for future therapy, but for now we should maximize the opportunities offered by surgery. In this article, we will examine recent updates in understanding of brain tumour biology and the implications for management. We will look at the latest research on brain tumour diagnosis, and then focus on management of the three most common brain tumours e glioma, meningiomas and cerebral metastases. We examine the surgical approach and adjuncts that should be employed to optimize patient outcomes.
Keywords Cerebral metastases; glioma; meningioma; surgery
Introduction Surgical oncology benefits patients across many cancer types, including brain tumours. For patients diagnosed with brain tumours, surgery is nevertheless rarely if ever curative. Uncertainties therefore remain as to the optimal surgical management for an individual patient. Surgery may be offered to aid control of tumour growth, but brain function is not localized into neatly defined anatomical regions, so attempts at tumour control need to be balanced against risk of operative morbidity. Surgical biopsy as a minimum is essential to obtaining a histological diagnosis, which informs adjuvant therapy options. Surgical tumour debulking may also be desirable to alleviate patient symptoms or control raised intracranial pressure. Surgical decision making needs to be informed by a thorough understanding of the available surgical and non-surgical treatment strategies, the likely pathology of a patient’s tumour and how surgical and adjuvant therapies will impact on the natural history of the tumour. In this article we will explore the latest thinking on brain tumour management, focusing on the most commonly operated brain tumour, glioma.
Summary of 2016 World Health Organization classification of tumours of the central nervous system C C C C C C C
Classification of CNS tumours
C
Most central nervous system tumours occur in the brain. Some rarer tumours affect the spinal cord and may be included in the differential diagnosis of relevant neurological symptoms. We focus here on tumours that affect the brain in adults. Paediatric tumours are fortunately rare, but no less devastating than adult
C C C C C C C C
Paul M Brennan PhD FRCS (SN) is Senior Clinical Lecturer and Honorary Consultant Neurosurgeon, Centre for Clinical Brain Sciences, Cancer Research UK Edinburgh Centre, UK. Conflicts of interest: none declared.
SURGERY 36:11
C
Diffuse astrocytic and oligodendroglial tumours Other astrocytic tumours Ependymal tumours Other gliomas Choroid plexus tumours Neuronal and mixed neuronaleglial tumours Tumours of the Pineal region Embyronal tumours Tumours of cranial and paraspinal nerves Meningiomas Mesenchymal, non-meningothelial tumours Melanocytic tumours Lymphomas Histiocytic tumours Germ cell tumours Tumours of the sellar region Metastatic tumours
Box 1
630
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
tumours have both oligodendroglial and astroglial components, so are called oligoastrocytomas. This is problematic because it belies a lack of understanding about the actual tumour biology. Determination of tumour type and grade (the aggressiveness of the tumour) from histological analysis of formalin-fixed tissue alone is prone to poor inter- and intra-observer agreement, with only 36e62% and 51e74%, respectively, in one study.3 The 2016 classification provides us with a more objective way of sub-classifying some brain tumours, which should lead to less inter-observer variation and more reproducibility. While the basic tumour categories remain the same, targeted molecular data adds additional value to the histological data, permitting stratification of patient tumours in a more objective and reliable way, reflecting our knowledge about tumour biology. The molecular stigmata included in the classification were chosen based on the demonstration in clinical practice that they are correlated with prognosis. The most significant impact of the molecular stratification is probably in glioma, the most common primary brain tumour. We will examine the molecular features in more detail when we discuss glioma more generally. The field of molecular pathology will continue to evolve as discovery science expands our understanding of tumour biology. Even since the 2016 classification update a new and innovative classification system has been reported based on genome-wide DNA methylation profiles of brain tumours.4 This approach has already identified new subclasses of glioma that were previously unknown, providing further opportunities to reappraise the way that we diagnose and manage patients. The expectation is that molecular stratification of brain tumours will drive identification of novel and more effective therapies, as well as describing the subgroups of patients most likely to benefit. There remains much though that we do not understand about the biology of tumours and it may be that multiple tumour, immune and microenvironment features need to be integrated into a single classification system before a tumour can be categorized in a way that reliable predicts treatment responsiveness and patient outcome.
pathology in a patient with a previous diagnosis of a malignancy elsewhere in the body, a secondary tumour may be suspected. Diagnosis of a possible brain tumour is particularly challenging in primary care. Guidelines, for example from the National Institute of Health and Care Excellence (https://www.nice. org.uk/guidance/ng12), provide advice that MRI brain imaging should be considered in any patient with ‘with progressive, subacute loss of central neurological function.’ This is quite nonspecific advice, and unsurprisingly only approximately 1% of patients referred for open/direct access brain imaging from primary to exclude a brain tumour have a brain tumour. More complex symptom combinations may be able to support identification of people who should be prioritized for urgent brain imaging. Researchers are also examining whether a liquid biopsy might be able to identify brain tumours even before brain imaging. Various strategies are being explored. Purification and analysis of small amounts of cell-free tumour DNA in the blood is technically challenging, particularly it seems with brain tumours, where the evidence suggests there is little cell free tumour DNA in the peripheral circulation. DNA-based strategies also depend on having identified which specific gene(s) is/are most likely to be sensitive and specific for the tumour. A second strategy is to use infrared spectroscopy to obtain a spectral profile of all the molecules in a patient sample, not just the DNA. Researchers have demonstrated that with this approach bloods samples from patients with brain tumours can be distinguished from those without brain tumours with a greater than 80% specificity and sensitivity.2,5,6 A serum blood test that reports high sensitivity and specificity for a brain tumour diagnosis may offer the possibility of supporting doctors in primary care to triage people most at risk of a brain tumour for urgent imaging. The downside of this approach is not yet knowing what molecules are responsible for the tumour-specific spectral patterns. Both of these liquid biopsies are the subject of ongoing clinical studies.
Cranial imaging Where a tumour is suspected, brain imaging can rapidly exclude or validate the diagnosis. Imaging should be arranged urgently, because even patients who appear relatively well can quickly deteriorate. The brain may accommodate a rise in intracranial pressure caused by tumour growth, but the compensatory mechanisms can be suddenly overwhelmed by a relatively small further change in the size of the lesion, caused for example by haemorrhage into the tumour, leading to rapid decompensation, brain herniation and death. Seizures can also be fatal in patients with raised intracranial pressures. Debate exists as to whether a normal CT is adequate to exclude a brain tumour. The NICE guidelines of suspected cancer recognition and referral (updated 2017) recommend consideration of an urgent direct access to MRI scan of the brain (or CT scan if MRI is contraindicated). A rapid CT may though be more use to the patient if they otherwise need to wait a few weeks for a MRI. A simple non-contrast CT scan may suffice to confirm or refute a suspicion of a brain tumour, particularly for malignant tumours. With a modern high-resolution CT scanner, few tumours are missed. Intravenous contrast will provide more detail of a tumour and help determine the tumour type. Non-contrast
Diagnosing a brain tumour Where a tumour is in or adjacent to an eloquent brain area, the patient may present with a weakness or seizure. The symptoms of a tumour can nevertheless be very non-specific, including personality change, memory loss or headache. Doctors and patients often worry that headache may be an indicator of tumour, and it is the most common symptom in patients with a brain tumour. However, headache is much more commonly not associated with a brain tumour and the likelihood of any individual person’s headache being caused by a brain tumour is very low. The difficulty in diagnosis leads to delays in referral for brain imaging. Brain tumour patients see their GP on average five times before a tumour is diagnosed. Many patients end up being diagnosed via the emergency department when their sequence of non-specific symptoms is followed by a weakness, dysphasia, seizure or just an inability to manage. There are no individual symptoms that are sufficiently sensitive or specific to be able to clearly identify patients at risk of having a brain tumour, so timely diagnosis depends on the treating physician having a suspicion of a possible tumour diagnosis. Where there are new symptoms of intracranial
SURGERY 36:11
631
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
Figure 1 Classical MRI appearance of GBM on T1 sequence after gadolinium contrast administration. GBMs are intrinsic to the brain, enhance heterogeneously with contrast and are usually localized to a single region in the brain.
Figure 3 Classical MRI appearance of cerebral metastases on T1 sequence after gadolinium contrast administration. The radiological appearance of brain metastases can vary. Classically, multiple foci of tumours, in discrete brain regions, located at the grey white matter junction, are suggestive of metastases.
CT scans can be misinterpreted, for example where there has been haemorrhage into an underlying tumour, or where oedema from a tumour is mistaken for stroke. Where urgent MRI imaging is not available, rapid access to a CT should be adequate to exclude a large mass lesion, with contrast or MRI imaging arranged in follow-up for where uncertainty exists, or simply to characterize the nature of the tumour. MRI provides more detail and is crucial for both surgical and radiotherapy planning. It is usually possible to make a provisional diagnosis of tumour type from CT and MR imaging. This is important to permit discussion with the patient and at the tumour multidisciplinary team (MDT) meeting as to the treatment choices. Glioblastoma multiformes (GBMs) are intrinsic to the brain, enhance heterogeneously with contrast and are usually localized to a single focus in the brain (Figure 1). Meningiomas are adjacent to the cerebral meninges. A cerebrospinal fluid (CSF) cleft is often seen between the tumour and cortex of the brain, and the tumour itself enhances homogeneously with contrast (Figure 2).
The radiological appearance of brain metastases can vary. Classically, multiple foci of tumours, in discrete brain regions, located at the greyewhite matter junction, are suggestive of metastases (Figure 3). They can have various patterns of contrast enhancement. A single brain metastasis may be easily confused with a GBM. A single dural based metastasis may be difficult to differentiate from a meningioma. If the brain tumour may be a secondary tumour from primary disease elsewhere on the body then a whole-body CT may be recommended to assess the extent of the primary disease and to ascertain whether there is a more suitable location than the brain for tumour biopsy.
Surgical decision making Although brain imaging can provide a good prediction of tumour type, a formal histological examination of tissue is needed to confirm the diagnosis and to provide the important molecular details. The priority then is for a tumour biopsy. If the tumour location permits, then the surgical approach may be designed to permit more extensive debulking of the tumour, rather than just biopsy. This may be desirable if reducing the bulk of the tumour is likely to reduce symptoms, or prevent an acute deterioration in symptoms. The MDT may consist of surgeons, pathologists, radiologists, oncologists, neurologists, specialist nurses, administrators, data managers and others. The MDT is central to discussion of clinical cases at each stage in a patient’s journey. We will look at the three most common tumours: glioma, meningioma and secondary brain tumours. We will examine the surgical treatment options, the other treatments available, and the prognosis for patients.
Primary brain tumours The two main types of primary brain tumour are glioma and meningioma. In operative surgical practice, gliomas are the most common. Many meningiomas are found incidentally and do not require surgery. The most common and most malignant glioma is glioblastoma.
Figure 2 Classical MRI appearance of meningioma on T1 sequence after gadolinium contrast administration. Meningiomas are adjacent to the cerebral meninges, in this case the superior sagittal sinus. A CSF cleft is often seen between the tumour and cortex of the brain, and the tumour itself enhances homogeneously with contrast.
SURGERY 36:11
632
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
The correlation of these genetic mutations with outcome has helped define clinical patient subtypes. The challenge for the neuro-oncology community is to develop a suite of treatment options that is bespoke for these different molecular subgroups. Currently, the paucity of effective therapies (and the limited efficacy of the ones that we do have) limits the impact of the additional information added by molecular subtyping. It nevertheless provides an opportunity to begin to target specific cohorts of patients into clinical trials with a view to developing more bespoke treatment regimens.
Gliomas One-third of the primary brain tumours diagnosed each year are glioma. GBM is the most common, affecting 4 per 100,000 of the UK population annually. The peak incidence of GBM is 63 years of age, but they can affect people of any age. Unlike some other solid tumours where known factors contribute to risk of tumour development, we do not know of any specific risk factors for glioma. Only very few people develop tumours as part of inherited conditions such as neurofibromatosis or Li Fraumeni syndrome. Gliomas were historically thought to arise from differentiated glial cells (the support cells of the brain’s nervous system) and the nomenclature that developed to subcategorize these tumours related to how the constituent tumour cells appeared to recapitulate features of different glial cells, as we discussed earlier. These different histological tumour types predicted clinical behaviour and treatment responsiveness to some degree. The most malignant glioma, glioblastoma multiforme, was so called for its lack of clearly defined features. There has been a change in our understanding at the cellular level of how gliomas develop. The model where tumour development follows acquisition of a mutation in a differentiated glioma cell has been replaced by the so-called cancer stem cell hypothesis.7 Here a subgroup of cells within the glioma that have characteristics similar to normal neural stem cells are thought to be responsible for tumour development and treatment resistance. These stem-like cells can divide to give rise to more cancer stem cells, but also to the differentiated cells that form the bulk of the tumour. The stem celllike behaviour of these tumour cells is changing the way that we are thinking about drug development and targeting. By modelling these cells in the laboratory, we may be better able to predict which compounds will be effective in the clinic. The 2016 revised WHO classification system for brain tumours has been particularly relevant to gliomas. It builds on the astrocytoma / oligodendroglioma / mixed oligo-astrocytoma / glioblastoma nomenclature by assessing genetic alterations in IDH1 or 2 genes, the ATRX gene, 1p and 19q chromosomes. Isocitrate dehydrogenase 1 (IDH1) was discovered through a genome-wide association study to be an important determinant of outcome in people with gliomas.8 A mutation in the IDH1 gene is associated with improved survival. The histological classification of glioma and glioblastoma is therefore supplemented with assessment of IDH1 status. There are several possible IDH1 mutations, but the most common is a single point mutation in codon 132. An antibody raised against the mutant protein can be used in routine immunohistochemistry to easily determine its presence. Other mutations of IDH1 and mutations in IDH2 can be identified through gene sequencing. ATRX is a regulator of chromatin remodelling and transcription, although its precise role in gliomas is not known. It is most important in distinguishing two subtypes of IDH mutant cells. IDH mutant tumours with mutant ATRX are astrocytomas, whereas IDH mutant tumours with 1p19q co deletion are oligodendrogliomas. Deletion of both the short arm chromosome 1 (1p) and the long arm of chromosome 19 (19q) is indicative of a better prognosis for patients, and occurs in a subset of patients with IDH mutant gliomas. Patients with IDH mutation, but without 1p19q, should be screened for TERT and ATRX mutations.
SURGERY 36:11
In 2005, a landmark trial published in the NEJM defined the standard of care therapy for people diagnosed with GBM.9 The so-called Stupp protocol recommends surgical resection, followed by radiotherapy and concomitant Temozolomide. Craniotomy: the process of accessing the cranium to remove, debulk or biopsy a tumour is similar for most tumour types. The location of the tumour, depth within the brain and surgical goals will influence positioning of the patient and the precise details of the surgical techniques. Detailed discussions of these techniques are presented in more specialist texts. Gliomas are intrinsic to the matter of the brain. There are no boundaries at the edge of the tumour, no capsule to limit tumour spread. We know from histological studies of post mortem tissue that tumour cells can be found within the hemisphere of the brain contralateral to high-grade primary tumours. Tumour cells spread along white matter tracts. Even where the location of the tumour permits an attempt to debulk most the tumour tissue, it is not possible to remove all the tumour. Most high grade tumours recur within 2e3cm of the resection cavity margin. Where possible, maximum safe debulking of a high-grade glioma appears to give patients the best average survival. When a patient with a high grade-glioma has a post-contrast MRI scan, there is usually a peripheral contrast enhancement and it is this boundary that surgeons aim to resect. Beyond this boundary there will still be tumour cells and so-called supramaximal resection may confer a further survival advantage, but it must be weighed against the increased risk of neurological deficit. There is a debate as to whether low-grade gliomas should be biopsied or debulked. Where they are in an easily accessible and non-eloquent brain location, maximum safe excision seems sensible. Often though, these-low grade tumours are in locations where attempts at maximal excision risk neurological deficit. Some surgeons argue that neurological deficits are acceptable, because with time and rehabilitation the deficit can recover. Others argue that a biopsy is enough to give a diagnosis and that larger debulking surgery can be delayed until there is evidence of tumour transformation. A large cohort study reported in Norway suggests that debulking nevertheless has a significant survival advantage.10 Surgical adjuncts: there are several different surgical adjuncts to help surgeons maximize the (safe) extent of resection. Preoperative brain imaging can be loaded into a work station (e.g. BrainLabÒ or StealthÒ) that creates a three-dimensional image that can then be correlated to the patient’s head in real time, using optical or electromagnetic tracking technologies. Optical tracking uses reflective markers on a reference
633
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
instrument placed close to the head of the patient. Reflective markers are also placed on a pointer that the surgeon can use intraoperatively, and the location of the pointer is tracked with an infrared camera connected to the work station. Using image guidance, the surgeon can confirm the location of the tumour, identify critical adjacent structures and follow the debulking of the tumour to ensure maximum resection. A T1 post contrast MRI can be fused with diffusion tractor imaging and functional MRI sequences to navigate key white matter tracts and eloquent brain regions, respectively. The imaging is only a supplement to the surgeon’s own anatomical expertise. As the tumour is debulked, the brain will deform and the preoperative imaging will become less accurate by comparison to the ‘live’ brain. The degree of inaccuracy depends on the tumour location and size. Ultrasound may help the surgeon to assess the amount of residual tumour in real-time, and these ultrasound images can be used to re-calibrate the preoperative imaging. Intraoperative MRI (iMRI) is increasingly available and may be of benefit to assess the extent of tumour resection, particularly in low-grade gliomas and paediatric tumours. Here the anaesthetized patient can be imaged during surgery to assess the extent of tumour resection. The surgeon can then continue the case. iMRI certainly enhances the extent of tumour resected. Whether this improves patient outcomes remains to be demonstrated in prospective, randomized trials. Most tumour surgeons will perform at least part of the tumour surgery using a microscope to enhance illumination and magnification of the operative space. Tumour-specific fluorescence visible through the microscope can be used to further enhance the extent of surgery. 5-Aminolaevulinic acid is a metabolite of the haem degradation pathway that accumulates preferentially in glioma cells and is metabolized to a fluorescent metabolite, protoporphyrin IX. Excitation with 405 nm wavelength blue light results in observable fluorescence. The ‘pink’ tumour can be debulked and use of this strategy is associated with demonstrably larger extent of resection than with white light surgery alone.11 Unfortunately, the Stummer study was not powered for survival and we do not know for certain that this enhanced surgery was associated with increased survival.11 The sensitivity and specificity of 5-ALA for glioma varies between tumours, and both false negatives and positives occur. While it is a valuable surgical adjunct, it needs to be complemented by other strategies to maximize safe surgical resection; just because tumour fluoresces does not mean that it is safe to remove. Other, more tumour-specific fluorescent dyes are under development.
low-risk anaesthetic candidate and needs to anticipate being able to tolerate being awake during surgery. Often the patient will remain awake once the tumour has been resected and while the craniotomy is closed. Many surgeons will perform the surgical approach and cranial opening with the patient asleep, waking the patient once the dura mater has been opened to expose the cortex. Incising the dura is uncomfortable. Once awake, the patient’s head remains fixed in position in pins, but they can speak and interact with the speech and language therapists, who can test multiple components of their language function. Power can also be assessed. To assess function in the exposed brain the surgeon stimulates the brain with a bipolar forceps. If the patient is, for example, naming objects when the brain is stimulated, and temporarily ceases to be able to correctly name an object, then the stimulated part of the brain is inferred to be responsible for that function. In this way, the eloquent cortex can be mapped. Bi-/multilingualism can be considered and the different sites of language mapped. Ultimately it is then up to the operating surgeon to make an assessment as to the relative risk to benefit ratio of proceeding with resection in the eloquent region. This will likely have been discussed with the patient beforehand. The motor cortex can also be mapped in the awake patient, although there is an increased risk of motor seizure. Seizure in an awake patient whose head is held in pins can be unpleasant and dangerous. The seizure can often be terminated by applying cold saline to the cortex of the brain. Seizures can also occur with stimulation of language areas. An alternative strategy is to assess motor function in an asleep patient using evoked motor potentials. Here the neurophysiologist places electrodes in specific muscle groups preoperatively that map across the motor homunculus of the primary motor cortex. Once the cranial vault is open, the surgeon can directly stimulate the brain while the neurophysiologist detects whether the stimulation has resulted in muscle activation. Visual and somatosensory evoked potentials are also possible. If a patient develops an impairment while awake or during intraoperative monitoring, the expectation is that if the resection in that region stops immediately then the deficit may recover. Adjuvant therapy: After surgery, tumour pathology will guide discussion about the role for each individual patient of further therapy. The degree to which molecular characteristics of a glioma influence treatment decisions depends on the grade of the tumour and the patient’s functional status. For patients with a high-grade glioma, there is unfortunately a paucity of treatment options and the recommendation for most patients will be radiotherapy with concomitant temozolomide (Stupp protocol). Temozolomide is an orally administered alkylating agent that damages DNA. The total radiotherapy dose is fractionated over several weeks to maximize the impact on the tumour while minimizing the side effects on normal brain. Once the course of radiotherapy has been completed, the monthly temozolomide continues for a total of six cycles. For many patients, the myelosuppressive side effect of chemotherapy delays the start of subsequent cycles and may limit the total number of cycles received. Methylation of the O6-methylguanine-DNA-methyltransferase (MGMT) promoter has been proposed as a prognostic marker of
Functional monitoring: image guidance, intraoperative imaging and direct tumour imaging with fluorescent markers do not provide any real-time information to the surgeon on the underlying function of the brain. When the surgical resection is in eloquent brain the surgeon may want to be able to assess the function of the tumour-infiltrated brain that is being removed intraoperatively, providing reassurance as to the proximity of important neural structures. The choice of strategy depends on the brain function that needs to be monitored and the preferences of patient and surgeon. During an ‘awake craniotomy’ the patient is quite literally awake while the tumour resection is proceeding. For an awake craniotomy, patient selection is critical. The patient needs to be a
SURGERY 36:11
634
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
array applied to the shaved cranium using adhesive patches, powered by a battery pack, for up to 18 hours a day. Other researchers are investigating the use of focused ultrasound to target the tumour, or to open the blood brain barrier in the region of the tumour, allowing drugs to enter the tumour microenvironment that would normally not be able to gain entry.
response to temozolomide. Temozolomide causes alkylation of the tumour genome, preventing cell division. MGMT can repair this damage. Methylation of MGMT suppresses transcription of the gene, thereby preventing repair of the damaged tumour genome. While this mechanism sounds elegant, in clinical practice the role of MGMT methylation in predicting treatment responsiveness and patient prognosis is not as clear cut. In the randomized clinical trial that defined the Stupp protocol, average survival following the trial intervention for patients with GBM was 14.6 months, compared to 12.1 months in the control group. Rarely has a 2 month increase in survival been so feted. Unfortunately, the situation is even worse when we consider all patients with GBM, not just the group eligible for a clinical trial. Median survival overall for patients in England with a GBM between 2007 and 2011 was just 6.1 months.12 The 1, 2and 5-year survivals rates were 28.4%, 11.5% and 3.4%, respectively. Understanding why a small group of patients with GBM survive more than 5 years may help identify strategies to better treat the bulk GBM population. For patients with low-grade glioma, the prognosis can usually be measured in single digit years rather than months. For most if not all patients, the low-grade tumour will undergo malignant transformation to a high-grade glioma. Patients with IDH wild type low-grade tumours will progress most rapidly.
Meningiomas Meningiomas are probably the most common primary brain tumour, but most do not require surgery. They arise from the arachnoid cap cells of the dural meninges that surround the brain. Most meningiomas appear sporadically, although previous cranial irradiation is a risk factor. Multiple meningiomata occur in neurofibromatosis type 2. Spontaneous mutations in the NF2 gene are also reported to be associated with meningiomas. Although meningiomas are less common in the surgical operating suite than gliomas, they are more common in the population, occurring at a rate of at least 8 per 1,000,000; only a quarter of patients are symptomatic at presentation. The increased availability of high-quality brain imaging has led to a surge in the incidence of people with so-called incidental meningiomas in the clinic. Differentiating an incidental lesion from a symptomatic lesion can be difficult if the patient has been imaged for non-specific symptoms. A lesion can often be considered incidental if it is small, located adjacent to noneloquent brain, and heavily calcified; the ectopic calcification suggests a long time in situ. The expectation is that an incidental lesion will not cause the patient trouble over their lifetime. Size is not necessarily a determinant of whether a meningioma is symptomatic or not. A small tumour located over the motor cortex could be more symptomatic than a larger tumour compressing the frontal lobe. Where the tumour is associated with brain oedema, consideration should be given to the need to excise it sooner rather than later. Little is known about the natural history of these incidental lesions discovered apparently by chance. Consequently, many patients will end up being imaged serially to look for a change in size. Whether a change in size precipitates a decision to operate will depend on the tumour location, the age of the patient and, above all, the patient’s preference. All these decisions need to be made with the assumption that the presumptive radiological diagnosis of a meningioma is correct. The diagnosis can only be validated with histological analysis of tumour tissue. There are rarer tumours with a similar radiological appearance to meningiomas, which need to be considered in the differential diagnosis. Meningiomas can be broadly divided into three grades. The higher the tumour grade, the greater the risk of recurrence. Recurrence risk is also influenced by the extent of surgical resection. Ideally the whole tumour and affected meninges will be removed, but this is not always technically possible, for example where the dura forms the wall of a dural venous sinus. With a risk of recurrence for even grade 1 tumours, and the potential morbidity from primary and subsequent surgeries, the classification of meningiomas as ‘benign’ belies the significant harm that patients can suffer. To obtain a histological diagnosis, surgery is recommended when the tumour grows significantly in size, or the patient is symptomatic. Surgery should maximize safe resection. Where the histology is reported as a grade 3 lesion, the oncologists
Tumour recurrence/progression: when gliomas recur/progress, it is important the patient is discussed at the MDT. Where possible, the surgical team will usually be asked to consider further surgery. Low-grade tumours often progress over several years, but high-grade tumours do so over weeks or months. Whether there is an advantage to surgery for recurrent tumour remains uncertain. However, the second-line chemotherapy options are very limited. A patient may be re-challenged with temozolomide if tumour progression has occurred a long time after the first block of therapy, but most likely they will be offered PCV (procarbazine, lomustine, vincristine), which has limited efficacy in glioma and significant toxicity. There has been no shortage of clinical trials to test new compounds in the recurrent high grade glioma population, but none have provided the significant benefit necessary to adopt them into routine clinical practice. This probably reflects our failure to adequately recapitulate tumour biology in the pre-clinical models of glioma used to screen compounds. It is certainly unlikely that surgery is going to be a definitive treatment for recurrent tumour. Hopefully, therefore, improved preclinical drug testing will identify new compounds that will be effective in the clinic. Other approaches are also being developed. Infiltrating immune cells account for a large proportion of the cells in a tumour. It is assumed that the tumour somehow manipulates the immune system to support its growth and to prevent its eradication. It may then be possible to harness the immune system to treat the tumour. This is a major focus of ongoing research and while no therapies have yet made it into routine clinical practice, there are many exciting approaches in development. Another approach has been to target the tumour noninvasively with electrical forces. These so-called tumour treating fields (TTFs) are thought to disrupt tumour cell division or cause cell death.13 The TTFs are administered via a transducer
SURGERY 36:11
635
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
would normally recommend radiotherapy before the tumour recurs. If the tumour is grade 2, it is not certain whether radiotherapy should be performed before or after recurrence is noted, and so the ROAM (Radiation versus Observation following surgical resection of Atypical Meningioma) trial was developed.14 There is no chemotherapy agent in use for routine management of meningiomas. The identification of a molecular pathway associated with tumour development and the NF2 gene has heralded the promise of a molecularly targeted therapy, but this has yet to emerge in the clinic.
REFERENCES 1 Louis DN, Perry A, Reifenberger G, et al. The 2016 World Health Organization classification of tumors of the central nervous system: a summary. Acta Neuropathol 2016; 131: 803e20. 2 Louis DN, Ohgaki H, Wiestler OD, et al. The 2007 WHO classification of tumours of the central nervous system. Acta Neuropathol 2007; 114: 97e109. 3 Mittler MA, Walters BC, Stopa EG. Observer reliability in histological grading of astrocytoma stereotactic biopsies. J Neurosurg 1996; 85: 1091e4. Available: http://www.ncbi.nlm.nih.gov/ pubmed/8929500. 4 Capper D, Jones DTW, Sill M, et al. DNA methylation-based classification of central nervous system tumours. Nature 2018; 555: 469e74. Available: http://www.nature.com/doifinder/10. 1038/nature26000. 5 Hands JR, Clemens G, Stables R, et al. Brain tumour differentiation: rapid stratified serum diagnostics via attenuated total reflection Fourier-transform infrared spectroscopy. J Neurooncol 2016; 127: 463e72. 6 Smith BR, Ashton KM, Brodbelt A, et al. Combining random forest and 2D correlation analysis to identify serum spectral signatures for neuro-oncology. Analyst 2016; 141: 3668e78. Available: http:// xlink.rsc.org/?DOI¼C5AN02452H. 7 Singh SK, Clarke ID, Terasaki M, et al. Identification of a cancer stem cell in human brain tumors. Cancer Res 2003; 63: 5821e8. 8 Yan H, Parsons DW, Jin G, et al. Mutations in gliomas. N Engl J Med 2009; 360: 765e73. 9 Stupp R, Mason WP, van den Bent MJ, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med 2005; 352: 987e96. Available: http://www.ncbi.nlm.nih. gov/pubmed/15758009. 10 Jakola AS, Myrmel KS, Kloster R, et al. Comparison of a strategy favoring early surgical resection vs a strategy favoring watchful waiting in low-grade gliomas. JAMA J Am Med Assoc 2012; 308: 1881e8. 11 Stummer W, Pichlmeier U, Meinel T, Wiestler OD, Zanella F, Reulen H-J. Fluorescence-guided surgery with 5-aminolevulinic acid for resection of malignant glioma: a randomised controlled multicentre phase III trial. Lancet Oncol 2006; 7: 392e401. Available: http://www.sciencedirect.com/science/article/pii/ S1470204506706659. 12 Brodbelt A, Greenberg D, Winters T, Williams M, Vernon S, Collins VP. Glioblastoma in England: 2007e2011. Eur J Cancer 2015; 51: 533e42. 13 Stupp R, Taphoorn M, Driven L, et al. Tumor treating fields (TTFields)dA novel cancer treatment modality: translating preclinical evidence and engineering into a survival benefit with delayed decline in quality of life. Int J Radiat Oncol Biol Phys 2017; 99: 1316. Available: http://ovidsp.ovid.com/ovidweb.cgi? T¼JS&CSC¼Y&NEWS¼N&PAGE¼fulltext&D¼emexb& AN¼619752156. 14 Jenkinson MD, Javadpour M, Haylock BJ, et al. The ROAM/EORTC-1308 trial: radiation versus Observation following surgical resection of Atypical Meningioma: study protocol for a randomised controlled trial. Trials 2015; 16.
Secondary brain tumours The most common tumours to metastasize to the brain are breast, renal, melanoma and colorectal primaries, although more rarely metastasis from tumours such as prostate and oesophageal tumours can occur. The management of the primary cancers is crucial to outcome in patients with brain metastases, but is beyond the scope of this article. Most patients with metastatic brain tumours will not be referred to neurosurgery. Where the systemic tumour is not controlled, the brain lesions are very small, or multiple, there may be little added benefit of surgery. Conversely, if the disease is well controlled outside the brain and there is a single symptomatic lesion then surgery is an option to discuss with the patient. Surgery may relieve an obstruction to CSF flow causing hydrocephalus. Even when the systemic disease it not controlled, or there are multiple brain lesions, there may be a role for excision of a symptomatic lesion. Surgery to obtain a histological diagnosis is indicated where the brain lesion(s) has the radiological appearance of a metastatic tumour but no primary disease has been previously identified or diagnosed. Surgery must be focused on minimizing any additional functional morbidity. Metastasis is evidence of systemic dissemination of the tumour and the benefit of surgery may be symptomatic and not in terms of overall survival. Similar to glioma surgery, there is support for supra-maximal resection of metastatic lesions, to remove cells outside the main bulk of the tumour, where this will not contribute to neurological morbidity. Following surgery, there may be a role for additional cranial radiotherapy to try to prevent recurrence. Whole brain radiotherapy can be associated with significant cognitive morbidity, so radiation may be targeted as stereotactic radiosurgery (SRS) to the tumour bed. SRS may also be used to treat small lesions not suitable for surgical resection. Cranial surgery for management of metastasis should ideally be undertaken in the context of a plan for additional systemic therapy.
Conclusion While the drive to ‘cure’ cancer often seems to be focused on innovations in chemotherapeutics, surgery has and will remain at the core of brain tumour management. Surgeons may not be able to cure cancer, but surgery delivers where the quest for novel therapies continues be disappoint. Developments in surgical technologies should focus on enhancing real-time surgical decision making and to maximize safe extent of resection. A
SURGERY 36:11
636
Ó 2018 Elsevier Ltd. All rights reserved.
Tumours of the central nervous system
tumours. There are some similarities in the tumour types that affect adults and children, but the differences have also taught us much about the biology and aetiology of these tumours. For one paediatric tumour, medulloblastoma, progress in our understanding of the tumour’s molecular phenotype has pioneered an approach to personalized medicine where the molecular phenotype impacts on treatment and prognosis. A better understanding of the biology of other brain tumours may also help us to improve stratification of affected patients for therapy, which in turn could inform the allocation of different treatments with improved outcomes. We are not there yet, but insights into tumour biology have already driven an updated system of tumour classification that is improving on traditional histological diagnosis. Tumours in the brain may be either primary or secondary tumours. Primary brain tumours are those that develop from cells in the brain. Secondary, or metastatic, tumours are those that have spread from elsewhere in the body. The most common tumours that metastasize to the brain are breast, renal, melanoma and colorectal tumours. There are 17 broad categories of primary CNS tumour according to the WHO 2007 classification, each with multiple subcategories. However, only a handful of the possible tumour types account for most of the tumours that are seen in clinical and surgical practice. The most common primary brain tumours are glioma and meningioma. In 2016 the World Health Organization (WHO) classification of tumours of the CNS was updated according to a new molecular nomenclature.1 This superseded the 2007 classification that was based on histological tumour features. The histology descriptors are based on similarities between tumour cells and the appearance of normal brain cells.1,2 (Box 1). For example, a glioma whose constituent cells look like astrocytes will be an astrocytoma, but if the cells look like oligodendrocytes then is it an oligodendroglioma. The problem arises though that some
Paul M Brennan
Abstract Surgery plays a key role in the management of brain tumours. The prognosis for a brain tumour diagnosis remains poor, but without the input of the surgical team, the outcomes would be bleaker. Innovations in understanding of the molecular biology of brain tumours may identify new targets for future therapy, but for now we should maximize the opportunities offered by surgery. In this article, we will examine recent updates in understanding of brain tumour biology and the implications for management. We will look at the latest research on brain tumour diagnosis, and then focus on management of the three most common brain tumours e glioma, meningiomas and cerebral metastases. We examine the surgical approach and adjuncts that should be employed to optimize patient outcomes.
Keywords Cerebral metastases; glioma; meningioma; surgery
Introduction Surgical oncology benefits patients across many cancer types, including brain tumours. For patients diagnosed with brain tumours, surgery is nevertheless rarely if ever curative. Uncertainties therefore remain as to the optimal surgical management for an individual patient. Surgery may be offered to aid control of tumour growth, but brain function is not localized into neatly defined anatomical regions, so attempts at tumour control need to be balanced against risk of operative morbidity. Surgical biopsy as a minimum is essential to obtaining a histological diagnosis, which informs adjuvant therapy options. Surgical tumour debulking may also be desirable to alleviate patient symptoms or control raised intracranial pressure. Surgical decision making needs to be informed by a thorough understanding of the available surgical and non-surgical treatment strategies, the likely pathology of a patient’s tumour and how surgical and adjuvant therapies will impact on the natural history of the tumour. In this article we will explore the latest thinking on brain tumour management, focusing on the most commonly operated brain tumour, glioma.
Summary of 2016 World Health Organization classification of tumours of the central nervous system C C C C C C C
Classification of CNS tumours
C
Most central nervous system tumours occur in the brain. Some rarer tumours affect the spinal cord and may be included in the differential diagnosis of relevant neurological symptoms. We focus here on tumours that affect the brain in adults. Paediatric tumours are fortunately rare, but no less devastating than adult
C C C C C C C C
Paul M Brennan PhD FRCS (SN) is Senior Clinical Lecturer and Honorary Consultant Neurosurgeon, Centre for Clinical Brain Sciences, Cancer Research UK Edinburgh Centre, UK. Conflicts of interest: none declared.
SURGERY 36:11
C
Diffuse astrocytic and oligodendroglial tumours Other astrocytic tumours Ependymal tumours Other gliomas Choroid plexus tumours Neuronal and mixed neuronaleglial tumours Tumours of the Pineal region Embyronal tumours Tumours of cranial and paraspinal nerves Meningiomas Mesenchymal, non-meningothelial tumours Melanocytic tumours Lymphomas Histiocytic tumours Germ cell tumours Tumours of the sellar region Metastatic tumours
Box 1
630
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
tumours have both oligodendroglial and astroglial components, so are called oligoastrocytomas. This is problematic because it belies a lack of understanding about the actual tumour biology. Determination of tumour type and grade (the aggressiveness of the tumour) from histological analysis of formalin-fixed tissue alone is prone to poor inter- and intra-observer agreement, with only 36e62% and 51e74%, respectively, in one study.3 The 2016 classification provides us with a more objective way of sub-classifying some brain tumours, which should lead to less inter-observer variation and more reproducibility. While the basic tumour categories remain the same, targeted molecular data adds additional value to the histological data, permitting stratification of patient tumours in a more objective and reliable way, reflecting our knowledge about tumour biology. The molecular stigmata included in the classification were chosen based on the demonstration in clinical practice that they are correlated with prognosis. The most significant impact of the molecular stratification is probably in glioma, the most common primary brain tumour. We will examine the molecular features in more detail when we discuss glioma more generally. The field of molecular pathology will continue to evolve as discovery science expands our understanding of tumour biology. Even since the 2016 classification update a new and innovative classification system has been reported based on genome-wide DNA methylation profiles of brain tumours.4 This approach has already identified new subclasses of glioma that were previously unknown, providing further opportunities to reappraise the way that we diagnose and manage patients. The expectation is that molecular stratification of brain tumours will drive identification of novel and more effective therapies, as well as describing the subgroups of patients most likely to benefit. There remains much though that we do not understand about the biology of tumours and it may be that multiple tumour, immune and microenvironment features need to be integrated into a single classification system before a tumour can be categorized in a way that reliable predicts treatment responsiveness and patient outcome.
pathology in a patient with a previous diagnosis of a malignancy elsewhere in the body, a secondary tumour may be suspected. Diagnosis of a possible brain tumour is particularly challenging in primary care. Guidelines, for example from the National Institute of Health and Care Excellence (https://www.nice. org.uk/guidance/ng12), provide advice that MRI brain imaging should be considered in any patient with ‘with progressive, subacute loss of central neurological function.’ This is quite nonspecific advice, and unsurprisingly only approximately 1% of patients referred for open/direct access brain imaging from primary to exclude a brain tumour have a brain tumour. More complex symptom combinations may be able to support identification of people who should be prioritized for urgent brain imaging. Researchers are also examining whether a liquid biopsy might be able to identify brain tumours even before brain imaging. Various strategies are being explored. Purification and analysis of small amounts of cell-free tumour DNA in the blood is technically challenging, particularly it seems with brain tumours, where the evidence suggests there is little cell free tumour DNA in the peripheral circulation. DNA-based strategies also depend on having identified which specific gene(s) is/are most likely to be sensitive and specific for the tumour. A second strategy is to use infrared spectroscopy to obtain a spectral profile of all the molecules in a patient sample, not just the DNA. Researchers have demonstrated that with this approach bloods samples from patients with brain tumours can be distinguished from those without brain tumours with a greater than 80% specificity and sensitivity.2,5,6 A serum blood test that reports high sensitivity and specificity for a brain tumour diagnosis may offer the possibility of supporting doctors in primary care to triage people most at risk of a brain tumour for urgent imaging. The downside of this approach is not yet knowing what molecules are responsible for the tumour-specific spectral patterns. Both of these liquid biopsies are the subject of ongoing clinical studies.
Cranial imaging Where a tumour is suspected, brain imaging can rapidly exclude or validate the diagnosis. Imaging should be arranged urgently, because even patients who appear relatively well can quickly deteriorate. The brain may accommodate a rise in intracranial pressure caused by tumour growth, but the compensatory mechanisms can be suddenly overwhelmed by a relatively small further change in the size of the lesion, caused for example by haemorrhage into the tumour, leading to rapid decompensation, brain herniation and death. Seizures can also be fatal in patients with raised intracranial pressures. Debate exists as to whether a normal CT is adequate to exclude a brain tumour. The NICE guidelines of suspected cancer recognition and referral (updated 2017) recommend consideration of an urgent direct access to MRI scan of the brain (or CT scan if MRI is contraindicated). A rapid CT may though be more use to the patient if they otherwise need to wait a few weeks for a MRI. A simple non-contrast CT scan may suffice to confirm or refute a suspicion of a brain tumour, particularly for malignant tumours. With a modern high-resolution CT scanner, few tumours are missed. Intravenous contrast will provide more detail of a tumour and help determine the tumour type. Non-contrast
Diagnosing a brain tumour Where a tumour is in or adjacent to an eloquent brain area, the patient may present with a weakness or seizure. The symptoms of a tumour can nevertheless be very non-specific, including personality change, memory loss or headache. Doctors and patients often worry that headache may be an indicator of tumour, and it is the most common symptom in patients with a brain tumour. However, headache is much more commonly not associated with a brain tumour and the likelihood of any individual person’s headache being caused by a brain tumour is very low. The difficulty in diagnosis leads to delays in referral for brain imaging. Brain tumour patients see their GP on average five times before a tumour is diagnosed. Many patients end up being diagnosed via the emergency department when their sequence of non-specific symptoms is followed by a weakness, dysphasia, seizure or just an inability to manage. There are no individual symptoms that are sufficiently sensitive or specific to be able to clearly identify patients at risk of having a brain tumour, so timely diagnosis depends on the treating physician having a suspicion of a possible tumour diagnosis. Where there are new symptoms of intracranial
SURGERY 36:11
631
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
Figure 1 Classical MRI appearance of GBM on T1 sequence after gadolinium contrast administration. GBMs are intrinsic to the brain, enhance heterogeneously with contrast and are usually localized to a single region in the brain.
Figure 3 Classical MRI appearance of cerebral metastases on T1 sequence after gadolinium contrast administration. The radiological appearance of brain metastases can vary. Classically, multiple foci of tumours, in discrete brain regions, located at the grey white matter junction, are suggestive of metastases.
CT scans can be misinterpreted, for example where there has been haemorrhage into an underlying tumour, or where oedema from a tumour is mistaken for stroke. Where urgent MRI imaging is not available, rapid access to a CT should be adequate to exclude a large mass lesion, with contrast or MRI imaging arranged in follow-up for where uncertainty exists, or simply to characterize the nature of the tumour. MRI provides more detail and is crucial for both surgical and radiotherapy planning. It is usually possible to make a provisional diagnosis of tumour type from CT and MR imaging. This is important to permit discussion with the patient and at the tumour multidisciplinary team (MDT) meeting as to the treatment choices. Glioblastoma multiformes (GBMs) are intrinsic to the brain, enhance heterogeneously with contrast and are usually localized to a single focus in the brain (Figure 1). Meningiomas are adjacent to the cerebral meninges. A cerebrospinal fluid (CSF) cleft is often seen between the tumour and cortex of the brain, and the tumour itself enhances homogeneously with contrast (Figure 2).
The radiological appearance of brain metastases can vary. Classically, multiple foci of tumours, in discrete brain regions, located at the greyewhite matter junction, are suggestive of metastases (Figure 3). They can have various patterns of contrast enhancement. A single brain metastasis may be easily confused with a GBM. A single dural based metastasis may be difficult to differentiate from a meningioma. If the brain tumour may be a secondary tumour from primary disease elsewhere on the body then a whole-body CT may be recommended to assess the extent of the primary disease and to ascertain whether there is a more suitable location than the brain for tumour biopsy.
Surgical decision making Although brain imaging can provide a good prediction of tumour type, a formal histological examination of tissue is needed to confirm the diagnosis and to provide the important molecular details. The priority then is for a tumour biopsy. If the tumour location permits, then the surgical approach may be designed to permit more extensive debulking of the tumour, rather than just biopsy. This may be desirable if reducing the bulk of the tumour is likely to reduce symptoms, or prevent an acute deterioration in symptoms. The MDT may consist of surgeons, pathologists, radiologists, oncologists, neurologists, specialist nurses, administrators, data managers and others. The MDT is central to discussion of clinical cases at each stage in a patient’s journey. We will look at the three most common tumours: glioma, meningioma and secondary brain tumours. We will examine the surgical treatment options, the other treatments available, and the prognosis for patients.
Primary brain tumours The two main types of primary brain tumour are glioma and meningioma. In operative surgical practice, gliomas are the most common. Many meningiomas are found incidentally and do not require surgery. The most common and most malignant glioma is glioblastoma.
Figure 2 Classical MRI appearance of meningioma on T1 sequence after gadolinium contrast administration. Meningiomas are adjacent to the cerebral meninges, in this case the superior sagittal sinus. A CSF cleft is often seen between the tumour and cortex of the brain, and the tumour itself enhances homogeneously with contrast.
SURGERY 36:11
632
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
The correlation of these genetic mutations with outcome has helped define clinical patient subtypes. The challenge for the neuro-oncology community is to develop a suite of treatment options that is bespoke for these different molecular subgroups. Currently, the paucity of effective therapies (and the limited efficacy of the ones that we do have) limits the impact of the additional information added by molecular subtyping. It nevertheless provides an opportunity to begin to target specific cohorts of patients into clinical trials with a view to developing more bespoke treatment regimens.
Gliomas One-third of the primary brain tumours diagnosed each year are glioma. GBM is the most common, affecting 4 per 100,000 of the UK population annually. The peak incidence of GBM is 63 years of age, but they can affect people of any age. Unlike some other solid tumours where known factors contribute to risk of tumour development, we do not know of any specific risk factors for glioma. Only very few people develop tumours as part of inherited conditions such as neurofibromatosis or Li Fraumeni syndrome. Gliomas were historically thought to arise from differentiated glial cells (the support cells of the brain’s nervous system) and the nomenclature that developed to subcategorize these tumours related to how the constituent tumour cells appeared to recapitulate features of different glial cells, as we discussed earlier. These different histological tumour types predicted clinical behaviour and treatment responsiveness to some degree. The most malignant glioma, glioblastoma multiforme, was so called for its lack of clearly defined features. There has been a change in our understanding at the cellular level of how gliomas develop. The model where tumour development follows acquisition of a mutation in a differentiated glioma cell has been replaced by the so-called cancer stem cell hypothesis.7 Here a subgroup of cells within the glioma that have characteristics similar to normal neural stem cells are thought to be responsible for tumour development and treatment resistance. These stem-like cells can divide to give rise to more cancer stem cells, but also to the differentiated cells that form the bulk of the tumour. The stem celllike behaviour of these tumour cells is changing the way that we are thinking about drug development and targeting. By modelling these cells in the laboratory, we may be better able to predict which compounds will be effective in the clinic. The 2016 revised WHO classification system for brain tumours has been particularly relevant to gliomas. It builds on the astrocytoma / oligodendroglioma / mixed oligo-astrocytoma / glioblastoma nomenclature by assessing genetic alterations in IDH1 or 2 genes, the ATRX gene, 1p and 19q chromosomes. Isocitrate dehydrogenase 1 (IDH1) was discovered through a genome-wide association study to be an important determinant of outcome in people with gliomas.8 A mutation in the IDH1 gene is associated with improved survival. The histological classification of glioma and glioblastoma is therefore supplemented with assessment of IDH1 status. There are several possible IDH1 mutations, but the most common is a single point mutation in codon 132. An antibody raised against the mutant protein can be used in routine immunohistochemistry to easily determine its presence. Other mutations of IDH1 and mutations in IDH2 can be identified through gene sequencing. ATRX is a regulator of chromatin remodelling and transcription, although its precise role in gliomas is not known. It is most important in distinguishing two subtypes of IDH mutant cells. IDH mutant tumours with mutant ATRX are astrocytomas, whereas IDH mutant tumours with 1p19q co deletion are oligodendrogliomas. Deletion of both the short arm chromosome 1 (1p) and the long arm of chromosome 19 (19q) is indicative of a better prognosis for patients, and occurs in a subset of patients with IDH mutant gliomas. Patients with IDH mutation, but without 1p19q, should be screened for TERT and ATRX mutations.
SURGERY 36:11
In 2005, a landmark trial published in the NEJM defined the standard of care therapy for people diagnosed with GBM.9 The so-called Stupp protocol recommends surgical resection, followed by radiotherapy and concomitant Temozolomide. Craniotomy: the process of accessing the cranium to remove, debulk or biopsy a tumour is similar for most tumour types. The location of the tumour, depth within the brain and surgical goals will influence positioning of the patient and the precise details of the surgical techniques. Detailed discussions of these techniques are presented in more specialist texts. Gliomas are intrinsic to the matter of the brain. There are no boundaries at the edge of the tumour, no capsule to limit tumour spread. We know from histological studies of post mortem tissue that tumour cells can be found within the hemisphere of the brain contralateral to high-grade primary tumours. Tumour cells spread along white matter tracts. Even where the location of the tumour permits an attempt to debulk most the tumour tissue, it is not possible to remove all the tumour. Most high grade tumours recur within 2e3cm of the resection cavity margin. Where possible, maximum safe debulking of a high-grade glioma appears to give patients the best average survival. When a patient with a high grade-glioma has a post-contrast MRI scan, there is usually a peripheral contrast enhancement and it is this boundary that surgeons aim to resect. Beyond this boundary there will still be tumour cells and so-called supramaximal resection may confer a further survival advantage, but it must be weighed against the increased risk of neurological deficit. There is a debate as to whether low-grade gliomas should be biopsied or debulked. Where they are in an easily accessible and non-eloquent brain location, maximum safe excision seems sensible. Often though, these-low grade tumours are in locations where attempts at maximal excision risk neurological deficit. Some surgeons argue that neurological deficits are acceptable, because with time and rehabilitation the deficit can recover. Others argue that a biopsy is enough to give a diagnosis and that larger debulking surgery can be delayed until there is evidence of tumour transformation. A large cohort study reported in Norway suggests that debulking nevertheless has a significant survival advantage.10 Surgical adjuncts: there are several different surgical adjuncts to help surgeons maximize the (safe) extent of resection. Preoperative brain imaging can be loaded into a work station (e.g. BrainLabÒ or StealthÒ) that creates a three-dimensional image that can then be correlated to the patient’s head in real time, using optical or electromagnetic tracking technologies. Optical tracking uses reflective markers on a reference
633
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
instrument placed close to the head of the patient. Reflective markers are also placed on a pointer that the surgeon can use intraoperatively, and the location of the pointer is tracked with an infrared camera connected to the work station. Using image guidance, the surgeon can confirm the location of the tumour, identify critical adjacent structures and follow the debulking of the tumour to ensure maximum resection. A T1 post contrast MRI can be fused with diffusion tractor imaging and functional MRI sequences to navigate key white matter tracts and eloquent brain regions, respectively. The imaging is only a supplement to the surgeon’s own anatomical expertise. As the tumour is debulked, the brain will deform and the preoperative imaging will become less accurate by comparison to the ‘live’ brain. The degree of inaccuracy depends on the tumour location and size. Ultrasound may help the surgeon to assess the amount of residual tumour in real-time, and these ultrasound images can be used to re-calibrate the preoperative imaging. Intraoperative MRI (iMRI) is increasingly available and may be of benefit to assess the extent of tumour resection, particularly in low-grade gliomas and paediatric tumours. Here the anaesthetized patient can be imaged during surgery to assess the extent of tumour resection. The surgeon can then continue the case. iMRI certainly enhances the extent of tumour resected. Whether this improves patient outcomes remains to be demonstrated in prospective, randomized trials. Most tumour surgeons will perform at least part of the tumour surgery using a microscope to enhance illumination and magnification of the operative space. Tumour-specific fluorescence visible through the microscope can be used to further enhance the extent of surgery. 5-Aminolaevulinic acid is a metabolite of the haem degradation pathway that accumulates preferentially in glioma cells and is metabolized to a fluorescent metabolite, protoporphyrin IX. Excitation with 405 nm wavelength blue light results in observable fluorescence. The ‘pink’ tumour can be debulked and use of this strategy is associated with demonstrably larger extent of resection than with white light surgery alone.11 Unfortunately, the Stummer study was not powered for survival and we do not know for certain that this enhanced surgery was associated with increased survival.11 The sensitivity and specificity of 5-ALA for glioma varies between tumours, and both false negatives and positives occur. While it is a valuable surgical adjunct, it needs to be complemented by other strategies to maximize safe surgical resection; just because tumour fluoresces does not mean that it is safe to remove. Other, more tumour-specific fluorescent dyes are under development.
low-risk anaesthetic candidate and needs to anticipate being able to tolerate being awake during surgery. Often the patient will remain awake once the tumour has been resected and while the craniotomy is closed. Many surgeons will perform the surgical approach and cranial opening with the patient asleep, waking the patient once the dura mater has been opened to expose the cortex. Incising the dura is uncomfortable. Once awake, the patient’s head remains fixed in position in pins, but they can speak and interact with the speech and language therapists, who can test multiple components of their language function. Power can also be assessed. To assess function in the exposed brain the surgeon stimulates the brain with a bipolar forceps. If the patient is, for example, naming objects when the brain is stimulated, and temporarily ceases to be able to correctly name an object, then the stimulated part of the brain is inferred to be responsible for that function. In this way, the eloquent cortex can be mapped. Bi-/multilingualism can be considered and the different sites of language mapped. Ultimately it is then up to the operating surgeon to make an assessment as to the relative risk to benefit ratio of proceeding with resection in the eloquent region. This will likely have been discussed with the patient beforehand. The motor cortex can also be mapped in the awake patient, although there is an increased risk of motor seizure. Seizure in an awake patient whose head is held in pins can be unpleasant and dangerous. The seizure can often be terminated by applying cold saline to the cortex of the brain. Seizures can also occur with stimulation of language areas. An alternative strategy is to assess motor function in an asleep patient using evoked motor potentials. Here the neurophysiologist places electrodes in specific muscle groups preoperatively that map across the motor homunculus of the primary motor cortex. Once the cranial vault is open, the surgeon can directly stimulate the brain while the neurophysiologist detects whether the stimulation has resulted in muscle activation. Visual and somatosensory evoked potentials are also possible. If a patient develops an impairment while awake or during intraoperative monitoring, the expectation is that if the resection in that region stops immediately then the deficit may recover. Adjuvant therapy: After surgery, tumour pathology will guide discussion about the role for each individual patient of further therapy. The degree to which molecular characteristics of a glioma influence treatment decisions depends on the grade of the tumour and the patient’s functional status. For patients with a high-grade glioma, there is unfortunately a paucity of treatment options and the recommendation for most patients will be radiotherapy with concomitant temozolomide (Stupp protocol). Temozolomide is an orally administered alkylating agent that damages DNA. The total radiotherapy dose is fractionated over several weeks to maximize the impact on the tumour while minimizing the side effects on normal brain. Once the course of radiotherapy has been completed, the monthly temozolomide continues for a total of six cycles. For many patients, the myelosuppressive side effect of chemotherapy delays the start of subsequent cycles and may limit the total number of cycles received. Methylation of the O6-methylguanine-DNA-methyltransferase (MGMT) promoter has been proposed as a prognostic marker of
Functional monitoring: image guidance, intraoperative imaging and direct tumour imaging with fluorescent markers do not provide any real-time information to the surgeon on the underlying function of the brain. When the surgical resection is in eloquent brain the surgeon may want to be able to assess the function of the tumour-infiltrated brain that is being removed intraoperatively, providing reassurance as to the proximity of important neural structures. The choice of strategy depends on the brain function that needs to be monitored and the preferences of patient and surgeon. During an ‘awake craniotomy’ the patient is quite literally awake while the tumour resection is proceeding. For an awake craniotomy, patient selection is critical. The patient needs to be a
SURGERY 36:11
634
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
array applied to the shaved cranium using adhesive patches, powered by a battery pack, for up to 18 hours a day. Other researchers are investigating the use of focused ultrasound to target the tumour, or to open the blood brain barrier in the region of the tumour, allowing drugs to enter the tumour microenvironment that would normally not be able to gain entry.
response to temozolomide. Temozolomide causes alkylation of the tumour genome, preventing cell division. MGMT can repair this damage. Methylation of MGMT suppresses transcription of the gene, thereby preventing repair of the damaged tumour genome. While this mechanism sounds elegant, in clinical practice the role of MGMT methylation in predicting treatment responsiveness and patient prognosis is not as clear cut. In the randomized clinical trial that defined the Stupp protocol, average survival following the trial intervention for patients with GBM was 14.6 months, compared to 12.1 months in the control group. Rarely has a 2 month increase in survival been so feted. Unfortunately, the situation is even worse when we consider all patients with GBM, not just the group eligible for a clinical trial. Median survival overall for patients in England with a GBM between 2007 and 2011 was just 6.1 months.12 The 1, 2and 5-year survivals rates were 28.4%, 11.5% and 3.4%, respectively. Understanding why a small group of patients with GBM survive more than 5 years may help identify strategies to better treat the bulk GBM population. For patients with low-grade glioma, the prognosis can usually be measured in single digit years rather than months. For most if not all patients, the low-grade tumour will undergo malignant transformation to a high-grade glioma. Patients with IDH wild type low-grade tumours will progress most rapidly.
Meningiomas Meningiomas are probably the most common primary brain tumour, but most do not require surgery. They arise from the arachnoid cap cells of the dural meninges that surround the brain. Most meningiomas appear sporadically, although previous cranial irradiation is a risk factor. Multiple meningiomata occur in neurofibromatosis type 2. Spontaneous mutations in the NF2 gene are also reported to be associated with meningiomas. Although meningiomas are less common in the surgical operating suite than gliomas, they are more common in the population, occurring at a rate of at least 8 per 1,000,000; only a quarter of patients are symptomatic at presentation. The increased availability of high-quality brain imaging has led to a surge in the incidence of people with so-called incidental meningiomas in the clinic. Differentiating an incidental lesion from a symptomatic lesion can be difficult if the patient has been imaged for non-specific symptoms. A lesion can often be considered incidental if it is small, located adjacent to noneloquent brain, and heavily calcified; the ectopic calcification suggests a long time in situ. The expectation is that an incidental lesion will not cause the patient trouble over their lifetime. Size is not necessarily a determinant of whether a meningioma is symptomatic or not. A small tumour located over the motor cortex could be more symptomatic than a larger tumour compressing the frontal lobe. Where the tumour is associated with brain oedema, consideration should be given to the need to excise it sooner rather than later. Little is known about the natural history of these incidental lesions discovered apparently by chance. Consequently, many patients will end up being imaged serially to look for a change in size. Whether a change in size precipitates a decision to operate will depend on the tumour location, the age of the patient and, above all, the patient’s preference. All these decisions need to be made with the assumption that the presumptive radiological diagnosis of a meningioma is correct. The diagnosis can only be validated with histological analysis of tumour tissue. There are rarer tumours with a similar radiological appearance to meningiomas, which need to be considered in the differential diagnosis. Meningiomas can be broadly divided into three grades. The higher the tumour grade, the greater the risk of recurrence. Recurrence risk is also influenced by the extent of surgical resection. Ideally the whole tumour and affected meninges will be removed, but this is not always technically possible, for example where the dura forms the wall of a dural venous sinus. With a risk of recurrence for even grade 1 tumours, and the potential morbidity from primary and subsequent surgeries, the classification of meningiomas as ‘benign’ belies the significant harm that patients can suffer. To obtain a histological diagnosis, surgery is recommended when the tumour grows significantly in size, or the patient is symptomatic. Surgery should maximize safe resection. Where the histology is reported as a grade 3 lesion, the oncologists
Tumour recurrence/progression: when gliomas recur/progress, it is important the patient is discussed at the MDT. Where possible, the surgical team will usually be asked to consider further surgery. Low-grade tumours often progress over several years, but high-grade tumours do so over weeks or months. Whether there is an advantage to surgery for recurrent tumour remains uncertain. However, the second-line chemotherapy options are very limited. A patient may be re-challenged with temozolomide if tumour progression has occurred a long time after the first block of therapy, but most likely they will be offered PCV (procarbazine, lomustine, vincristine), which has limited efficacy in glioma and significant toxicity. There has been no shortage of clinical trials to test new compounds in the recurrent high grade glioma population, but none have provided the significant benefit necessary to adopt them into routine clinical practice. This probably reflects our failure to adequately recapitulate tumour biology in the pre-clinical models of glioma used to screen compounds. It is certainly unlikely that surgery is going to be a definitive treatment for recurrent tumour. Hopefully, therefore, improved preclinical drug testing will identify new compounds that will be effective in the clinic. Other approaches are also being developed. Infiltrating immune cells account for a large proportion of the cells in a tumour. It is assumed that the tumour somehow manipulates the immune system to support its growth and to prevent its eradication. It may then be possible to harness the immune system to treat the tumour. This is a major focus of ongoing research and while no therapies have yet made it into routine clinical practice, there are many exciting approaches in development. Another approach has been to target the tumour noninvasively with electrical forces. These so-called tumour treating fields (TTFs) are thought to disrupt tumour cell division or cause cell death.13 The TTFs are administered via a transducer
SURGERY 36:11
635
Ó 2018 Elsevier Ltd. All rights reserved.
NEUROSURGERY
would normally recommend radiotherapy before the tumour recurs. If the tumour is grade 2, it is not certain whether radiotherapy should be performed before or after recurrence is noted, and so the ROAM (Radiation versus Observation following surgical resection of Atypical Meningioma) trial was developed.14 There is no chemotherapy agent in use for routine management of meningiomas. The identification of a molecular pathway associated with tumour development and the NF2 gene has heralded the promise of a molecularly targeted therapy, but this has yet to emerge in the clinic.
REFERENCES 1 Louis DN, Perry A, Reifenberger G, et al. The 2016 World Health Organization classification of tumors of the central nervous system: a summary. Acta Neuropathol 2016; 131: 803e20. 2 Louis DN, Ohgaki H, Wiestler OD, et al. The 2007 WHO classification of tumours of the central nervous system. Acta Neuropathol 2007; 114: 97e109. 3 Mittler MA, Walters BC, Stopa EG. Observer reliability in histological grading of astrocytoma stereotactic biopsies. J Neurosurg 1996; 85: 1091e4. Available: http://www.ncbi.nlm.nih.gov/ pubmed/8929500. 4 Capper D, Jones DTW, Sill M, et al. DNA methylation-based classification of central nervous system tumours. Nature 2018; 555: 469e74. Available: http://www.nature.com/doifinder/10. 1038/nature26000. 5 Hands JR, Clemens G, Stables R, et al. Brain tumour differentiation: rapid stratified serum diagnostics via attenuated total reflection Fourier-transform infrared spectroscopy. J Neurooncol 2016; 127: 463e72. 6 Smith BR, Ashton KM, Brodbelt A, et al. Combining random forest and 2D correlation analysis to identify serum spectral signatures for neuro-oncology. Analyst 2016; 141: 3668e78. Available: http:// xlink.rsc.org/?DOI¼C5AN02452H. 7 Singh SK, Clarke ID, Terasaki M, et al. Identification of a cancer stem cell in human brain tumors. Cancer Res 2003; 63: 5821e8. 8 Yan H, Parsons DW, Jin G, et al. Mutations in gliomas. N Engl J Med 2009; 360: 765e73. 9 Stupp R, Mason WP, van den Bent MJ, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med 2005; 352: 987e96. Available: http://www.ncbi.nlm.nih. gov/pubmed/15758009. 10 Jakola AS, Myrmel KS, Kloster R, et al. Comparison of a strategy favoring early surgical resection vs a strategy favoring watchful waiting in low-grade gliomas. JAMA J Am Med Assoc 2012; 308: 1881e8. 11 Stummer W, Pichlmeier U, Meinel T, Wiestler OD, Zanella F, Reulen H-J. Fluorescence-guided surgery with 5-aminolevulinic acid for resection of malignant glioma: a randomised controlled multicentre phase III trial. Lancet Oncol 2006; 7: 392e401. Available: http://www.sciencedirect.com/science/article/pii/ S1470204506706659. 12 Brodbelt A, Greenberg D, Winters T, Williams M, Vernon S, Collins VP. Glioblastoma in England: 2007e2011. Eur J Cancer 2015; 51: 533e42. 13 Stupp R, Taphoorn M, Driven L, et al. Tumor treating fields (TTFields)dA novel cancer treatment modality: translating preclinical evidence and engineering into a survival benefit with delayed decline in quality of life. Int J Radiat Oncol Biol Phys 2017; 99: 1316. Available: http://ovidsp.ovid.com/ovidweb.cgi? T¼JS&CSC¼Y&NEWS¼N&PAGE¼fulltext&D¼emexb& AN¼619752156. 14 Jenkinson MD, Javadpour M, Haylock BJ, et al. The ROAM/EORTC-1308 trial: radiation versus Observation following surgical resection of Atypical Meningioma: study protocol for a randomised controlled trial. Trials 2015; 16.
Secondary brain tumours The most common tumours to metastasize to the brain are breast, renal, melanoma and colorectal primaries, although more rarely metastasis from tumours such as prostate and oesophageal tumours can occur. The management of the primary cancers is crucial to outcome in patients with brain metastases, but is beyond the scope of this article. Most patients with metastatic brain tumours will not be referred to neurosurgery. Where the systemic tumour is not controlled, the brain lesions are very small, or multiple, there may be little added benefit of surgery. Conversely, if the disease is well controlled outside the brain and there is a single symptomatic lesion then surgery is an option to discuss with the patient. Surgery may relieve an obstruction to CSF flow causing hydrocephalus. Even when the systemic disease it not controlled, or there are multiple brain lesions, there may be a role for excision of a symptomatic lesion. Surgery to obtain a histological diagnosis is indicated where the brain lesion(s) has the radiological appearance of a metastatic tumour but no primary disease has been previously identified or diagnosed. Surgery must be focused on minimizing any additional functional morbidity. Metastasis is evidence of systemic dissemination of the tumour and the benefit of surgery may be symptomatic and not in terms of overall survival. Similar to glioma surgery, there is support for supra-maximal resection of metastatic lesions, to remove cells outside the main bulk of the tumour, where this will not contribute to neurological morbidity. Following surgery, there may be a role for additional cranial radiotherapy to try to prevent recurrence. Whole brain radiotherapy can be associated with significant cognitive morbidity, so radiation may be targeted as stereotactic radiosurgery (SRS) to the tumour bed. SRS may also be used to treat small lesions not suitable for surgical resection. Cranial surgery for management of metastasis should ideally be undertaken in the context of a plan for additional systemic therapy.
Conclusion While the drive to ‘cure’ cancer often seems to be focused on innovations in chemotherapeutics, surgery has and will remain at the core of brain tumour management. Surgeons may not be able to cure cancer, but surgery delivers where the quest for novel therapies continues be disappoint. Developments in surgical technologies should focus on enhancing real-time surgical decision making and to maximize safe extent of resection. A
SURGERY 36:11
636
Ó 2018 Elsevier Ltd. All rights reserved.