Hi, my name is Daniel Orringer. I'm an associate professor of neurosurgery at NYU Grossman School of Medicine, and happy to present this presentation on simulated rhombohistology to eliminate guesswork during high-grade glioma surgery for the AANS 2020 course on Technologic Adjuncts for Malignant Glioma Surgery. Starting with my disclosures, I'm a shareholder and medical advisor to Adveneo Imaging Incorporated. I also work as a consultant for NX Development. So I'd like to start today by giving you a picture inside my OR. I love this picture because it gives the viewer a feeling for how central imaging is to the practice of neurosurgery. If you see all the screens that we have in our OR, you realize that we're looking at the field, we're looking at our screens, we're integrating imaging data into everything that we do. We have a navigation system, we have a 3D virtual visualization system, we're monitoring the patient's vital signs, we're visualizing the operative field with ultrasound, and of course, our eyes don't leave the patient for the vast majority of the operation. This picture underscores the importance of what we can see and how central that is to achieving good outcomes in brain tumor surgery. And I would contend that our results are only as good as the images that are used to guide us. There are two central questions that I want to explore today and talk to you about an innovative approach to addressing questions that are central to glioma surgery. First of all, taking a couple steps back, verifying that the patient that we've taken to the OR for an operation on a malignant glioma does indeed have that diagnosis. And that's something that can be quite difficult to detect based on conventional MRI. Our ability to differentiate high-grade glioma from other lesions is improving as preoperative MRI is improving, but we still need to verify the diagnosis in many cases, and that diagnosis may not be clear from any of the radiographic or gross factors that we have at the time of the operation. So a key element of making sure that we're doing the right operation is ensuring that we have the correct diagnosis. Establishing diagnosis is fundamental to do everything else that we consider important in the operations that we do for our high-grade glioma patients. And number two, given the challenges that we all have in differentiating normal brain from tumor-infiltrated tissue, it's understanding where that interface is between the tumor and the normal brain and how to thread that needle between taking out as much tumor as possible while leaving the normal brain unperturbed. The importance of diagnosis, and specifically histology, as a key element of decision-making in surgery dates back to the 20s and 30s when Dr. Cushing and his pathologist Louise Eisenhardt published a paper in the American Journal of Pathology arguing for the importance of the use of intraoperative histology to make clinical decisions during brain tumor resection. And today, it's widely accepted that histologic data are essential for formulating surgical strategy. However, the current state of the art for intraoperative histology requires sending tissue to a pathology lab either for smear or frozen section, both of which can be difficult to interpret. There can be delays associated with the process, and they can be difficult to use in an iterative fashion to inform surgical decision-making. As those who are viewing this presentation know, the existing workflow for intraoperative histology commonly requires a biopsy, passing the tissue off the field, transportation of that tissue to a frozen section lab, sectioning of that tissue by a technician. Sometimes that process is complicated by the fact that there may be multiple frozen sections, multiple specimens that need to be addressed by a single lab from different ORs, so there could be a queue of tissues waiting to be sectioned. Then the tissue, once it's sectioned, needs to be stained, and a pathologist needs to come in and render a diagnosis and then communicate that back to the operating room. This process is cumbersome, as I mentioned, and has evolved very little since Cushing's time. The inefficiencies of this process create a barrier to the use of microscopic data during an operation and make surgical decision-making quite detached from the process of histology. This actually is counter to what Dr. Cushing envisioned back in his time, when he felt that intraoperative histology should be integrated into our decision-making at each step of the surgical treatment of our patients. We've re-envisioned intraoperative histology through the implementation of a technique in optics. Without going into the physics at this point, I would contrast conventional histology and the complex logistics associated with that with this new technology called stimulated Raman histology, where a tissue is simply placed onto a slide, a cover slip is placed on the tissue, and the slide is then inserted into the imaging system, and a virtual histologic image is acquired in two to five minutes. The wait time for intraoperative diagnosis, which can be up to 30 or 45 minutes, can be reduced tenfold to just a matter of minutes. And as I mentioned, the workflow is really straightforward. Small samples are placed onto custom-designed slides, the slide is inserted in an imager, and the diagnostic image is returned in minutes. The steps of transporting tissue to the frozen section lab for smearing or sectioning, mounting, fixation, staining, and dehydration are no longer necessary. Also, SRH images, as I'll discuss, are natively digital, and the SRH imager can be configured to upload to a PAX, much in the way an MRI or CT scanner would, enabling remote interpretation by a pathologist. So what's behind this technology? I think it's best understood via an analogy. Just as MRI images are created by mapping NMR spectral data within an anatomical plane, Raman spectral data, which is determined by the vibrational properties of the chemical bonds, can be mapped onto a microscopic field of view. Both technologies create images in a non-destructive manner without the need for dyes or labels. However, Raman signals are quite weak, and it took 16 hours to generate this low-resolution Raman image with Raman microscopy. My colleague, Dr. Chris Freutiger, in 2008, described a technique called stimulated Raman scattering microscopy, which boosts Raman signal strength by 10,000-fold, enabling high-resolution microscopic imaging. And this paved the way for us to design a system that we could bring into the operating room. And once we did that, several years back, we realized that stimulated Raman histology captures all of the essential cytoarchitectural features that are required to differentiate normal and reactive brain from glial tumors and extra-axial tumors, non-glial tumors. And when we posed these images to a neuropathologist, we found that the accuracy for making a diagnosis was comparable using SRH images, much in the same way it would be using conventional histology. In our study, our accuracy exceeded 92% for predicting diagnosis in comparison to conventional histology, and these findings have been now validated in over 1,400 brain tumor patients. This is a picture from our OR, and what we now have is both the ability to look at tissue grossly and to look at it microscopically in the operating room. And what this means for us is that we can go through the process of verifying that we have a malignant glioma in the operating room with us based on histologic data rather than just guessing or relying on the cumbersome pathway of conventional histology to verify that diagnosis. So there's sort of a decision-making process that we all learn as residents that is required to get to a diagnosis of malignant glioma in the operating room and verify that we are indeed dealing with the pathology that we expect. And a lot of times this is sort of an automatic thing. This is the basic dance that we do. I get this idea of a box step from a great presentation by Dr. Lawton, and our box step kind of looks like this. Our first question is, is the tissue neoplastic or not? If it's not, we have to consider the possibilities that we're dealing with inflammation, infection, stroke, or something else. It's important as surgeons to be able to identify these conditions accurately and rapidly in the operating room because if we're in a position where one of these patients ends up on our operating room table, surgery is over once you make that diagnosis and medical treatment is most important. If we can confirm that we've got a neoplasm, we need to also verify that it's not something like lymphoma or germinoma that's best treated with medical means. If it is, again, the operation's over and that patient should be closed up and treated medically. If it's not a sort of medically treated tumor, we have to determine, is this something coming from the brain or is this something from outside of the brain? A lot of times we can see whether it's a meningioma, for example, or something intrinsic like a glioma, but in other cases that can be difficult. It's important to identify a metastasis or meningioma because the surgical goals and surgical approaches to these tumors are certainly different than that of high-grade glioma. If it is intrinsic, it's important to verify whether it's high-grade or low-grade because the decisions that we make have important implications for our patients. And there's debate about this, but if a patient has a very aggressive tumor in a highly eloquent area, we may be more conservative, understanding that quality of life may be more important than trying to get out every last tumor cells. We might try with a low-grade glioma where a patient's outcome can really be impacted in a major way based on the quality of our surgical resection. So for high-grade gliomas, glioblastoma, anaplastic gliomas, and PNAT, we're trying our best to get as much tumor as we possibly can, understanding that there are some limitations to what we can achieve with surgery and we want to be respectful of the patient's function. If it's low-grade, there's a different consideration, especially if it's something like a pyrocytic astrocytoma, ganglioglioma, or ependymoma, where these well-circumscribed gliomas can actually be cured surgically if they're fully removed. And in astrocytoma and oligodendroglioma, we're talking about the potential to add years of survival on to a patient's lives if we're able to get a very good resection. So being able to answer all of these questions very quickly is an expectation and sort of the standard of care, but very difficult to execute unless you have high-quality, rapid histologic data, and we can end up in difficult positions without this. So as surgeons, we like to be in control of everything that goes into all the decisions that are made to make sure we're doing the safest possible job. So I would contend that we should have control of the practice of neuropathology as it relates to our patient that we're operating on. So in the last couple of slides I'm going to go through, I'll explore a number of cases where we had a complex clinical presentation or unclear clinical presentation and use histology to be able to make sure that we're performing the correct operation. So this is a 57 year old gentleman with a 30 pack year history of smoking and a newly diagnosed right upper lung mass. He also had this intracranial lesion. The differential diagnosis for this lesion would include high grade glioma. Most likely given the lung mass, this could be a lung metastasis. It could also be any of the other inflammatory, infectious conditions. But given his age and given his smoking history, we all thought this was likely to be a lung metastasis. When we took this patient to the OR, we biopsied the tissue and noted that the wall of the lesion had this appearance. At the core of it, there was a dense neutrophilic infiltrate. The diagnosis in this case turned out to be a brain abscess. So this is a patient in whom you certainly do not want to be doing a gross total resection on. You want to be able to establish the diagnosis, evaluate the tissue that you've encountered, and recognize, okay, this is not neoplastic at all. This patient needs to be treated with drainage of what turned out to be an abscess, and then with medication. And we should not worry about trying to resect the wall of the lesion, as that can cause unnecessary damage. So this is one situation where we were quite fooled, based on the radiographic impression of the diagnosis in this patient's case when we took him to the operating room. So this is a bit more of a straightforward case, a 71-year-old woman who had a history of a wide local excision of a melanoma in her right arm, and presented with visual hallucinations. Here, the histology was as expected. This is a tumor, certainly hypercellular, with some pleomorphism, large, regular, high nuclear to cytoplasmic ratio cells that have very sharp margins, no evidence of axons within this. This clearly looks like a melanoma, and this is something that was very well circumscribed, and that we could take out fully, and treat with adjuvant immunotherapy. This one was a bit less straightforward. This was a 65-year-old gentleman with a history of immunosuppression who presented with this slightly diffusion-restricting right parietal mask. When we took this patient to the OR, there was evidence of perivascular cuffing, discohesive cells, cellular morphology that was round and monomorphic. The cells were very infiltrative. There was a high nuclear to cytoplasmic ratio. Another case, another slightly diffusion-restricting lesion that shows similar histology. These are nice examples of CNS lymphoma, and both situations in which aggressive resection is not indicated. Once we're able to get the tissue that we need for diagnosis the operation is done, and resection should not be carried out. So this is a good example of another situation where the histology guides our decision-making, and we don't want to be doing the operation that we would do on a high-grade glioma patient for a patient who has histology that looks like lymphoma. Another interesting case was this 32-year-old woman with new onset seizures and intermittent headaches who had this right temporal mask that was read by a radiologist as markedly enhancing and well-defined with a diagnosis that should include a BRAF tumor such as a PXA, so something that could fall into the category of high-grade glioma, certainly possible. The radiologist also said, oh, this could be a solitary fibrous tumor, metastasis. Essentially, we don't know. When we took her to the operating room, what we saw was histology that was very characteristic of meningioma. So we see here world cells with evidence of calcification and the formation of somoma bodies. And this was handled like a meningioma and looked like a meningioma microscopically and turned out to be a grade one meningioma that we could fully resect. So rather than confusing this for some kind of an infiltrative glioma, we were able to see that this is a well-circumscribed meningioma and that gross total resection is very important and feasible in this situation. So these are two tumors radiographically that were billed as gliomas and they look like gliomas based on the MR imaging. The question here is, are we able to differentiate between high and low grade? Both of these tumors were enhancing. One was in a 43-year-old. The one that turned out to be a low grade was in a 43-year-old. 70-year-old was the one that had a high-grade glioma. But this is just meant to show that SRH is valuable in being able to differentiate these extremely hypercellular high-grade gliomas with evidence of microvascular perforation from those that are hypotenormocellular with some evidence of cytolitosis and evidence of infiltration along axons that are characteristic of a low-grade glioma. And this differentiation between high-grade and low-grade is something that becomes pretty straightforward as you use this technique. And it can be used to guide our surgical management. As I mentioned before, differentiating amongst well-circumscribed and diffuse gliomas is very important. So these are two adults who are billed with high-grade glioma coming into the operating room. One of them has a highly infiltrative glioma in which their tumor cells intermix with these white streaky things that are axons. The other on the right side of the screen has a pylocytic astrocytoma with these dark things, which are Rosenthal fibers and a hair-like appearance of the tissue, which is why it's called pylocytic astrocytoma. Certainly a very different surgical approach when you're dealing with a diffuse glioma of the cerebellum compared to something like a pylocytic. In the case of the pylocytic, gross total resection should be curative. In the case of this infiltrative glioma, we wanna be extremely careful because surgery is not gonna cure this patient. They will need the addition of radiation and chemotherapy. And if this is close to the brainstem, it's likely infiltrative. So very different picture in our surgical decision-making depending on the histology. A situation where we commonly find ourselves in in the glioma surgery world is, are we dealing with viable tumor in our patient who has a history of glioma, or are we dealing with pseudoprogression? And the radiographic appearance is something that can be very difficult to use alone to differentiate these states. However, if we look at tissue from cases that have comparable radiographic appearances, contrast enhancing essentially necrotic or cystic lesions, we can see one in which there's clearly viable tumor with some treatment effect and another where we're just simply seeing gliosis and a lack of proliferating tumor. This has very important implications for how aggressive we wanna be during surgery. For a patient where we just see gliosis histologically, we wanna be relatively conservative as there's probably not much tumor burden. When we see recurrent tumor and we're there in the operating room, if it can be taken safely, it should be removed. So as I mentioned, the other side of this is that once we've established the proper diagnosis, the other essentially vexing problem that I know a lot of speakers in this series will touch on today is that we need to be able to differentiate between tumor and the adjacent neural brain and that's key to being able to achieve the best possible surgical results. Certainly our surgical training gives us the power to make good decisions the vast majority of the cases and generally as well-trained glioma surgeons, we're gonna be able to differentiate the tumor core from adjacent normal brain. At the margins though, we're somewhat limited in what we can do and there is a vast amount of data that suggests that as surgeons, we're generally able to get good results in most cases but achieving perfect results is very difficult even for the best glioma surgeons. So there's a suite of technology, much of what you'll hear about in other lectures in this series including fluorescence-guided surgery, intraoperative MRI, awake mapping, other neurophysiologic techniques and emerging optical techniques to be able to differentiate tumor from normal. All of these technologies though work on different scales and there is a way that we're able to employ stimulated Raman histology in order to be able to close the gap on what we can see with our eyes, what we can see with a 5-ALA, what we can see with intraoperative MRI and look on a microscopic scale at a resection cavity to determine the various levels of cellularity that we might encounter. In this situation of this 42-year-old woman with anaplastic oligodendroglioma co-deleted, we can see the varying histology depending on whether we're within the center of the lesion where we see an abundance of tumor cells infiltrating along axons. If we look at the inferior aspect of this lesion, we can see infiltrating tumor cells and at a grossly identical region of the cavity that was along the superior edge, the margin appears clean histologically. So it's still a work in progress to be able to identify the histologic endpoints that we should be going for, but we now have the technology with the capacity for rapid bedside diagnosis that allows us to differentiate amongst areas of the tumor where we've done a really good job taking out tumor and what we see left is histologically normal from those that might still require a bit of work. And especially in this tumor, which was in the frontal pole, we have the leeway to be able to make sure that all of our margins look essentially clean histologically and those that don't, we can go back and resect. So with this technology, we have the potential when deployed appropriately of achieving a microscopically complete result and pushing the limits as to what we can achieve. This is not gonna be feasible in all situations and with a very large tumor, this technology can be hard to monitor all areas of a large cavity, but I think it fits in nicely with some of the techniques we'll hear about in other parts of the series with MRI, with 5ALA. If you're concerned about a certain area within the tumor cavity, you can go and take a biopsy and be able to eliminate the guesswork associated with that region of the field and know definitively whether that region contains tumor cells or not. I'd like to end here and turn it over to the excellent panel of speakers that we have lined up today. And please reach out to me and let me know via email if there are any questions about the technology, if there's any clarifications I can make to this presentation. Thank you.

Dr. Daniel Orringer

simulated rhombohistology

high-grade glioma surgery

accurate diagnosis

intraoperative histology

brain tumors