Okay, so I want to thank the organizers for inviting me again to talk about the role of surgery and some of the advances of surgery in the management of childhood brain tumors. I have no relevant disclosures. I'm going to sort of hit the high points and talk about the most common childhood brain tumors, specifically low and high-grade gliomas, medulloblastomas, ependymomas, and diffuse intrinsic pontine gliomas. So one of the things to know is that brain tumors are the most common solid tumors in childhood, and they also have the highest mortality. Although there have been major improvements in the treatment of some types of childhood brain tumors, like low-grade gliomas and medulloblastomas, there are other subsets, like the high-grade gliomas and the brainstem gliomas, that continue to do very poorly. Extent of resection, which is one of the things that we can have an impact on as surgeons, is associated with outcome for many of the subtypes of childhood brain tumors. And in addition to the importance of debulking the tumor, we also provide specimens for molecular analysis, and this molecular data has really revolutionized how childhood brain tumors are considered and how they're managed. So I'll start off with low-grade gliomas. These are the most common supertentorial brain tumors, and the second most common infratentorial brain tumors. And these are broadly characterized as pylocytic tumors, which are often enhancing with a big cyst and often a mural nodule, and then non-pylocytic tumors, which are often non-enhancing and much more infiltrative. But for both types of tumors, resection extent has a major impact on outcome. These are some data published by Jeff Wisoff and colleagues from the Children's Cancer Group and the Pediatric Oncology Group, highlighting the fact that tumors that are amenable to gross total resection have about a 90-95% 10-year progression-free survival, whereas those that undergo incomplete resections do substantially worse. So for most hemispheric tumors of the supertentorial and infratentorial space, the goal is to try to get a gross total resection. This gets to be a little challenging, though, for the midline low-grade gliomas, those that involve the diencephalon. Some of these tumors are resectable. This is a unilateral thalamic tumor. It was a pylocytic astrocytoma. And we were able to remove it by a middle temporal gyrus approach. On the other hand, a tumor like this, which is a globular chiasmatic tumor that occurred in a child with NF, if I had shown the flare images, there would be signal tracking all the way back to the occipital lobes, goes all the way forward to the globes. And this is not a tumor that you can remove. And there probably isn't even a role for biopsy, because in kids with NF, you know that this is a low-grade glioma. One of the differences between childhood and adult low-grade gliomas is that the childhood tumors respond fairly well to chemotherapy. So even for tumors that are not resectable, like this one, also in a child with NF1, chemotherapy works very well. And this child was treated with carboplatinum and vincristin, one of the most common regimens used. And a year later, you can see the tumor is substantially smaller. And this child never required any additional treatment. Even big tumors, like this one, where we did an internal debulking, can respond very well to chemotherapy. This is the appearance 10 weeks after this postoperative MR, showing a dramatic shrinkage of the tumor. And this child ultimately remained progression-free for about 3 and 1⁄2 years. One of the downsides to chemotherapy is that, in most cases, it's not a cure. And a lot of these kids will progress anywhere from 3 to 5 years after they're started on conventional chemotherapy. So there's really a need for new approaches. It turns out there have been new ideas for management of low-grade gliomas, particularly pylocytic astrocytomas. Within the last 10 years, it's become apparent that most of these tumors are driven by alterations in the MAP kinase signaling pathway, which goes from the cell surface all the way down into the nucleus and promotes cell division, cell survival, and spreading. The most common mutations are a fusion involving the BRAF gene, which is up here, upstream of the MAP kinase pathway. They often can also have mutations of the BRAF gene. And then in kids with NF, they have loss of neurofibromin, which shifts the balance to the GTP form of BRAFs that then activates this pathway. So if you take all the pylocytic astrocytomas and look at them for these different molecular features, about 95% have some alteration in the MAP kinase pathway. And that's important because it's a druggable pathway. There are MAP kinase inhibitors, also BRAF inhibitors that have been tested in kids within the last several years, and they've all shown activity. These are results from one study conducted in the Pediatric Brain Tumor Consortium, the PBTC, using a MAP kinase inhibitor, selumetinib. And this is just a phase one study result, but what you'll notice in this waterfall plot, which shows changes in tumor size after treatment, is that the vast majority of these tumors have shrunk with this therapy, in many cases quite substantially. It turns out that those responses are durable. As long as you keep giving the drug, many of these kids will continue to respond for three or four years. So this then moved on to a phase two study, and now there is a phase three randomized trial that's comparing this agent versus a conventional chemotherapy regimen in the Children's Oncology Group. So this is really a paradigm shift in how these tumors are treated, using molecularly targeted therapy rather than conventional chemotherapy. Now I'm going to change gears and talk about high-grade gliomas. Historically, they were thought to have a better prognosis in kids than in adults, and it turns out that's wrong. There are different molecular subsets of tumors in kids than adults. The prognosis in the molecular subsets that overlap are actually virtually identical. And as with the low-grade gliomas, extent of resection has an impact on outcome. Not as dramatic, but these are some results from the Children's Cancer Group 945 study showing that tumors that are amenable to 90% resection or more have a much better outcome than those that aren't. And this may reflect not only the skill of the surgeon, but also where the tumor is located. So midline high-grade gliomas, as with midline low-grade gliomas, tend to do worse because they're not as resectable. Within the last 5 to 10 years, it's been noted that there are different molecular subgroups of high-grade gliomas. The first group that was identified by our lab was the IDH mutant subset. IDH mutations are seen in high-grade gliomas that occur in young adults. It turns out, probably not as a big surprise, that if you look at tumors that arise in teenagers, they also have a high incidence of IDH mutations, about 35%. They're rare in younger kids. And tumors that have IDH mutations do dramatically better than those that don't. Over the years, there have been other molecular subsets that have been identified. There are a high frequency of mutations in one of these histone genes, the H3.3 or 3.1 gene. And that comes in two varieties. If you have a mutation at the K27 locus, the patients do very poorly. They generally have midline tumors. The G34 locus, they do better. And those tumors often involve the hemispheres. There's also been a BRAF mutated subset identified, the IDH1 group that I mentioned earlier, and then several different subgroups that have amplification of tyrosine kinase receptors, like the adult high-grade gliomas. So it turns out that we now have individualized therapy within the children's oncology group for the BRAF subset, for the G34 subset, and for the K27 subset. We don't have any pediatric protocols for the IDH1 subgroup, but there are adult trials for that. And the importance of identifying these groups, not only does it influence what might be the best therapy, it has a major impact on outcome. As I showed you earlier, IDH1 tumors do well. These are some data from a European cooperative group study showing that IDH tumors do dramatically better than any of the other subsets. The K27 mutated tumors do terribly, and the G34 mutated tumors are intermediate. And then this is sort of like a wastebasket for all the other alterations that don't involve those three spots. So I will move on to medulloblastomas. In the past, these tumors were subdivided into average risk and high-risk groups, depending on the age of the patient, whether there are metastases, and whether there was a lot of bulky disease left behind after surgery. So this was up until about 2000, the way tumors were categorized. Then more recently, people have realized that the histology of the tumor has a big impact on outcome. So all medulloblastomas don't look identical. There are some that are very anaplastic-looking, some that are very desmoplastic-looking, and those groups do worse and better, respectively, than the classical, typical-looking medulloblastomas. And even more recently, the genomic profiles of these tumors have been identified. And there are at least four subgroups that are now incorporated within the World Health Organization classification. So they're defined by these different patterns of mutations, and I'll go through those in a minute. They're also subdivided based on their histology. And the treatments are different as well. As with the other two tumors I talked about, there's an impact of resection extent on outcome. These are old data from the Children's Oncology Group showing that tumors that had minimal residual disease did substantially better than those that had bulky residual disease. With the new molecular data, the group from Toronto, who's done large-scale analyses of medulloblastomas, demonstrated that the impact of resection extent is attenuated when you subdivide or stratify patients based on their molecular subgroups. But even after doing that, there still is a significant impact of resection extent. And so maximal safe resection remains the standard of care. So we want to take a tumor that looks like this and turn it into a tumor or a post-op scan that looks like that without hurting the brain in the process. That sometimes can be challenging because most of the tumor is not visible when you approach these lesions. So this is post-fossa craniotomy. And this is a nubbin of tumor in the cisterna magna between the cerebellar tonsils. And it's deceiving because this is the MRI scan. So the piece that you're seeing on that operative image is this little bit down here. And the rest of it is all up in the fourth ventricle and cerebellum. So the way I typically approach these is to define the plane between tumor and brain stem inferiorly so I have an idea about depth. And then I'll either elevate or incise the inferior part of the vermis and then core out the center of the tumor. And then define the plane between tumor and cerebellum. And then clear out tumor from within the fourth ventricle. And what you're left with if you've done an extensive resection is often a nice view of the fourth ventricle. This is the median RAFA and the aqueduct. And not uncommonly, there is a little area of tumor that has burrowed into the brain stem surface. We don't chase that because it has no impact on outcome. And it is associated with a lot of morbidity. So this is a tumor where a near total resection is usually the goal rather than a truly total resection. So after surgery, these kids nowadays are treated with radiation and chemotherapy. Before about 2000, tumors were treated pretty much exclusively with radiation. And it turns out that the results from radiation alone, which are shown here with two different doses, low dose and high dose radiation, are inferior to the combination of low dose radiation and chemotherapy. These are two different regimens that were tested in a randomized trial. So both of these regimens beat the results with radiation alone. So nowadays, the standard is so-called reduced dose radiation, 2340 centigrade of the neuroaxis, a boost to the posterior fossa, and then chemotherapy. And this is the most recent completed study for average risk medulloblastomas. That's those that are non-metastatic and extensively resected. And it attempted to reduce the craniospinal radiation dose even further to 1800 centigrade and also reduce the volume of the boost dose of radiation. Unfortunately, the group that got 1800 centigrade did worse than the group that got 2340. The boost volume had no impact on outcome. And this paper has yet to be published because molecular analysis are being done on all of these tumors, over 500 of them, to try to see if the negative effect of 1800 centigrade is influenced by whether the tumor is a sonic hedgehog tumor or a Wnt tumor or one of the other medulloblastoma pathways. The importance of recognizing these pathways, and this is a gene expression analysis, sort of giving you the idea that these tumors do cluster based on their expression of genes into at least four different groups. There's a group C, which does terribly, a group D, which is intermediate, a sonic hedgehog group, which has now been split into two groups that is also intermediate, and then the Wnt subgroup, which occurs in older kids, has a much better prognosis. There are strong age effects in these different subsets. So, sonic hedgehog tumors tend to occur in infants and older teenagers, whereas the Wnt group of tumors occurs in older kids, older than about eight. So, nowadays, the medulloblastomas are considered at least as four distinct subsets of tumors. This is a paper published by Michael Taylor several years ago showing a consensus classification scheme that was agreed to by neuro-oncologists around the world, but it turns out that things are a lot more complicated than that, and if you begin looking at these groups in more detail, you see that there are subsets that can be defined within these big groups. So, now, it's thought that there could be as many as 13 subsets of medulloblastomas. So, this is getting to be a little bit of a headache in terms of clinical trial design, because you can't really have trials for 13 different subsets of tumors, you know, with very small numbers of patients. So, they still are grouped using those four main subsets. So, we've had studies within the Pediatric Brain Tumor Consortium for the sonic hedgehog group. There's a study within the Children's Oncology Group for the Wnt subset, and then other investigational trials for group C tumors. The fourth tumor I'll talk about are ependymomas. These, more than any of the other subsets, have a major impact of resection extent on outcome, whereas there's a strong trend for the other tumors towards better outcome with extensive resection. For ependymomas, if you don't get a gross total resection, the likelihood of the child surviving long-term is very low, less than 30% and probably more like 10 to 20%. Whereas, if you're able to get a true total resection, the long-term survival rates are 60 to 70%. When these tumors come back, most of the time they come back locally. So, we tend to use involved field radiation rather than craniospinal radiation. So, that's the difference between these tumors and medulloblastomas. And the other difference from a therapy standpoint is that chemotherapy has not been shown to have an association with outcome in ependymomas. It was just examined in a phase three randomized trial with about 400 patients, but those results haven't been released yet. So, it's possible that this larger study will show an impact, but to date, nothing has been identified. So, the goal for these is, again, to try to get a gross total resection. And that can be challenging when you have a huge tumor in a young child. This was a two-year-old with this massive post-fossa tumor. And the way I approach these is by monitoring cranial nerve EMGs. Whatever I can monitor, I monitor. I also do BSCRs and SSEPs. Because the treacherous part of these tumors is taking them off the brain stem. The part on the surface shown here is often very well circumscribed. You can peel it right off the cerebellum. But along the brain stem, they're microscopically infiltrative. And you can certainly do some damage when you're trying to resect these. Fortunately, in this child, we were able to get a gross total resection. And she's done well long-term. This is a tumor type where second look surgery has also been shown to have a role. And this is an example of one case where that's what we pursued. This child presented with this large posterior fossa tumor. Turned out to be an ependymoma. And the child was an extremus. So, it was taken urgently to surgery. And we did a near total resection. There was tumor still within the medulla. And every time I tried to resect that little rind of tumor, the patient would become asystolic. So, I stopped. We then treated the child with chemotherapy for six weeks. You can see that the tumor looks like it's enhancing less. It's come out of the medulla to some extent. And then we went back in and did a gross total resection. So, this is actually incorporated in all the ependymoma clinical trials now. An option for second look surgery after a period of chemotherapy. As with the other tumors I've talked about, things have gotten a lot more complicated in classifying these tumors. So, we used to think of ependymomas as one group or two groups, anaplastic and non-anaplastic ependymomas. Now, it's recognized that there are nine subsets of ependymomas. People are beginning to think of treatment specifically targeted to these individual subsets. But to my knowledge, there have not been any trials launched for example, for posterior fossa A tumors or Rel A tumors. But I think that's coming. And then lastly, I'll talk about brainstem gliomas, both high grade and low grade. The diffuse brainstem gliomas like this, which involve the pons from front to back, have an absolutely horrible prognosis that has not changed in 30 years. These are results from a study in around 2000 showing a 20% one-year progression-free survival. The results from our current trials are not a bit better than this. So, there's definitely a need for innovative approaches. And people are pursuing that. Mark Sweedane at the Memorial Sloan-Kettering has done a number of studies with convection-enhanced delivery. And this is now moving to a multi-site trial within the Pediatric Brain Tumor Consortium. Other groups, including the Pacific Neuro-Oncology Consortium, have begun to stratify therapy based on molecular markers after biopsying these tumors. So, it used to be these tumors were never biopsied. Now, they are selectively being biopsied. Unfortunately, the results with targeted therapy so far have not been overwhelmingly positive. But I think as we get better agents, we may start combining them. We may start to see responses. And this is also a tumor type where immunotherapy has been used in a number of trials. So, there are options out there for these tumors. But we have a long way to go in terms of treating them better than we're doing right now. And I'd be remiss if I didn't mention that there are some groups of brainstem gliomas that aren't that malignant. Dorsally exophytic brainstem gliomas, cervical medullary tumors, and focal tumors have a much better prognosis. And they are potentially resectable. This is an example of one dorsally exophytic brainstem glioma that I did a gross total resection on years ago. And this child has never recurred. So, to summarize, I know I've covered a lot of territory very rapidly. But I wanted to give you an overview of the most important subgroups of childhood brain tumors. Surgery has an important role in managing tumors. It has a big impact on overall outcome. The appropriate treatment is determined both by the anatomy and the histology of the tumor. And some of this new molecular data is actually causing us to rethink some of our dogma in how patients are managed. Certainly in medulloblastomas and low-grade gliomas. And this molecular stratification and molecularly targeted treatment have revolutionized how postoperative care is done for several of these subsets of tumors. So this has been an exciting time to be a pediatric neurosurgeon or neuro-oncologist. There's a lot going on. And I think the field is going to continue changing over the next several years. And I'll be happy to take any questions. Thank you.

surgery

childhood brain tumors

extent of resection

targeted therapies

molecular analysis

research