Yale Interventional Oncology Lab

Transcript

• 00:00Thank you, thank you and welcome
• 00:03everyone to Cancer Center,
• 00:05grand rounds and really pleased with how
• 00:09we continue to innovate on formats in
• 00:12this new Zoom World of of the pandemic.
• 00:16And I think today's session is going
• 00:19to be interesting given the nature
• 00:22of of four speakers and various
• 00:25levels of their their training and
• 00:28professional development and are.
• 00:30Cancer Center in our University and
• 00:33health care system and you know,
• 00:35really focused on.
• 00:37The great work that's going on in
• 00:40radiology and biomedical imaging
• 00:42and the science associated with it,
• 00:45most notably in Interventional Oncology.
• 00:49Please introduce let us say the
• 00:52the leader for today's session.
• 00:54Doctor David Madoff is,
• 00:56as you may recall from last week,
• 00:59doctor Madoff is the Co director
• 01:01of Interventional Oncology of the
• 01:03Interventional and college research
• 01:05lab vice chair for clinical research,
• 01:08and the section chief for
• 01:10Interventional radiology.
• 01:11An Department of radiology
• 01:12and biomedical imaging,
• 01:14and David really has throughout his career
• 01:17have been an innovator in this space.
• 01:20And with his joining the faculty
• 01:22at Yale now a couple of years ago,
• 01:25I guess has really built out not only the
• 01:29capabilities in our clinical operations,
• 01:31but expanding research which
• 01:32is really exciting.
• 01:34And so David I turn it over to you to
• 01:37sort of introduce the other speakers
• 01:39and share with us all this great work.
• 01:44OK, thanks Charlie.
• 01:45So I'd like to again, thank you for giving
• 01:48me the opportunity to speak in Yelp,
• 01:50Cancer Center grand rounds.
• 01:51As you saw it last week I was part
• 01:54of a session on the multidisciplinary
• 01:56management of colorectal liver metastases.
• 01:58Today's session, as you stated it will
• 02:00be focused on Interventional oncology.
• 02:02What it is in some of the exciting
• 02:04research being done in our lab at Yale.
• 02:07Therefore, this program will focus on
• 02:09work being done by our trainees and
• 02:11not really on my own personal work.
• 02:13I will just be introducing the topics.
• 02:15And our vision and goals for the
• 02:18Interventional Oncology Program at Yale.
• 02:20So this session, as you may recall,
• 02:23was originally planned for March 10th
• 02:25and it was just after the kovid pandemic.
• 02:28So interesting, Lee,
• 02:29It's actually given us more time
• 02:32to make our data even more mature.
• 02:34So for the audience for today.
• 02:37So I'm really pleased. I think about that.
• 02:40So I'm going to start and then
• 02:42will be followed by Julia Shapiro,
• 02:45who happens to be the Co director
• 02:48of the intervention college.
• 02:49Live with me.
• 02:50He's actually an assistant professor
• 02:52of radiology and biomedical imaging,
• 02:55while actually,
• 02:56still,
• 02:56and I are resident so that shows you
• 02:59the level of achievement that some
• 03:02of our colleagues really have and
• 03:04then will be followed by Jessica
• 03:07Santana and tells Evie who are both
• 03:09graduate students in our lab and
• 03:12the work will all be talking about
• 03:14liver cancer and Interventional
• 03:16oncology related activities.
• 03:18So let me just share my screen.
• 03:27OK, so. Really,
• 03:31what is Interventional oncology?
• 03:32Well intermixed oncology is a
• 03:34subspecialty of Interventional
• 03:36radiology that utilizes minimally
• 03:37invasive image guided procedures
• 03:39to both diagnose and treat patients
• 03:41with various forms of cancer.
• 03:43The benefits of primary intervention
• 03:45oncology treatments or its
• 03:47immediate tumoricidal effects.
• 03:48As you I'm sure are aware,
• 03:51there minimally invasive,
• 03:52resulting in cost reductions
• 03:54and time Efficacy,
• 03:55as well as having minimal systemic side
• 03:58effects leading to an overall improved.
• 04:00Quality of life.
• 04:01The goal has been to make the case
• 04:05for us becoming the 4th pillar of
• 04:07cancer care and over the years
• 04:10I believe we have and this has
• 04:12been shown actually in many ways.
• 04:15These include having IO therapies
• 04:17incorporated into multiple NCCN
• 04:19guidelines which include you know
• 04:21colorectal metastases include
• 04:22HP be an even in renal cancer,
• 04:24having trials to assess the role of
• 04:27Percat Aneus management of cancer,
• 04:29and what I believe is most important.
• 04:32Is that,
• 04:33I believe now become a valued participator
• 04:36in most if not all of tumor boards.
• 04:40So what does it take to
• 04:42become a pillar of oncology?
• 04:44Clearly each clinical discipline needs
• 04:45to have a strong foundation in basic,
• 04:48translational and clinical research.
• 04:49We do have some work to do in this regard,
• 04:53but I must say I was really
• 04:55thrilled to see when I was in 2016
• 04:58as an invited faculty of Asco GI.
• 05:00My discipline next to my name was listed
• 05:02actually as an Interventional Oncologist,
• 05:04not an interventional radiologist,
• 05:06so I was really happy at that time,
• 05:08and I thought we actually made it.
• 05:12However,
• 05:12I really do believe that in terms of research
• 05:15and education of referring physicians,
• 05:17I think we're nearly there
• 05:18and on the right track,
• 05:20but I don't know if we've come there totally.
• 05:24So here you can see a long list of
• 05:28procedures that we do, and I owe.
• 05:31These can range from image guided biopsy,
• 05:34primary tumor therapy,
• 05:35palliative procedures,
• 05:36central venous access,
• 05:38an managing of complications that
• 05:40are either a result of cancer itself
• 05:44or related to cancer treatments.
• 05:46So I just wanted to highlight some
• 05:49challenging biopsies as this is one of
• 05:51the most important procedures that we do.
• 05:53Clearly it is difficult for Oncologist to
• 05:55treat patients in the absence of a diagnosis.
• 05:58That said,
• 05:59cases such as this one should be
• 06:01easily doable for any experienced
• 06:03interventional list,
• 06:03and as you can see,
• 06:05this patient has confirmed
• 06:08metastatic adenocarcinoma.
• 06:09So here is the second case,
• 06:11which I would say is much more
• 06:13challenging due to lack of any
• 06:15imaging finding on CT or ultrasound
• 06:16and you can see here it's closed
• 06:19proximity to the heart.
• 06:20Fortunately we were able to get
• 06:22adequate tissue sampling and help
• 06:24this patient get the appropriate
• 06:26chemotherapy that they needed to be on.
• 06:29So I just wanted to discuss a couple
• 06:31of primary intra interventional
• 06:32oncology tumor therapies,
• 06:34both of which will be discussed in
• 06:36research by our trainees later in
• 06:38the session here we have two more
• 06:40guided ablation or tumor
• 06:42image guided tumor ablation,
• 06:43so the goals of Ablation is to eradicate
• 06:46all valuable malignant cells and stem
• 06:48spare normal surrounding tissues,
• 06:50treat tissues with unfavorable location
• 06:51or pattern of distribution for resection
• 06:54and or have multiple comorbidities.
• 06:56These are the most often used in
• 06:58patients with low volume disease and
• 07:00required to bulking are typically
• 07:01done in an outpatient setting setting
• 07:04and these procedures are repeatable.
• 07:06So if we can see here we have
• 07:08radiofrequency ablation and microwave
• 07:10which are heat based therapies.
• 07:12We have Cryo Ablation which is a coal
• 07:14based therapy and we actually have
• 07:16irreversible electroporation which
• 07:17is really to electrocute the tumors.
• 07:19Believe it or not,
• 07:21by changing the ionic potentials
• 07:24between the between the membranes.
• 07:26So ablation of liver tumors,
• 07:28which is in this was its initial indication,
• 07:31as shown here in.
• 07:33This is 1 case of a patient with
• 07:36HTC that needed ablation as a
• 07:38bridge to transplant.
• 07:40As we can see, 2 1/2 years later,
• 07:43no tumor recurrence or any residual disease.
• 07:47Next was the case of an isolated
• 07:49colorectal liver metastasis.
• 07:50Who here you see a pet positive
• 07:53or hypermetabolic lesion in the
• 07:54right lobe of the liver who was
• 07:57successfully treated with Ablation
• 07:58and his team are free at one year?
• 08:01Follow up.
• 08:03This is a patient who,
• 08:05instead of having a liver lesion,
• 08:07has a lung nodule.
• 08:08Here we can see that based on
• 08:10having heat based thermal ablation,
• 08:12the patient did very well at seven months.
• 08:15As you can see, there's no residual
• 08:17tumor on the image Ng, and there was no
• 08:21local recurrence seen at three years.
• 08:24We can also treat bone metastasis.
• 08:26This is a patient with metastatic
• 08:28breast carcinoma with a focal right
• 08:30femoral and right hip pain limiting
• 08:32mobility for whatever reason,
• 08:34she refused radiation therapy
• 08:35and was treated with heat based
• 08:38thermal ablation and Samantha
• 08:39plasty is shown here and she had
• 08:42an immediate improvement in right
• 08:43femoral pain from 8 to 10 out of 10.
• 08:49This is a very interesting use of
• 08:52thermal ablation in this case,
• 08:53with cryoablation.
• 08:54This patient had metastatic Rectal
• 08:56Squamous Cell Carcinoma who had
• 08:58severe intractable 10 out of 10 pain,
• 09:00secondary to inoperable tumor recurrence
• 09:02infringing on the sacral nerve roots.
• 09:04She was successfully treated with
• 09:06partial cryo ablation of the bowel
• 09:08wall and you can see right here the
• 09:10low density ice she had immediate
• 09:13or should say complete resolution
• 09:15of her pain within about 3 days.
• 09:17So she had actually no.
• 09:19Team up until a year,
• 09:21at which time she actually died
• 09:23and then Lastly we were able
• 09:26to treat this liver tumor here.
• 09:28That was a budding the left
• 09:31hepatic bile duct.
• 09:32So we used Ayari again,
• 09:34that's basically electrocution,
• 09:35which is technically a non thermal ablation.
• 09:38So about nine months later the patient
• 09:41has no residual disease or recurrence.
• 09:45We also have more regional therapies
• 09:47such as transarterial embolization.
• 09:48In these cases are typically reserved
• 09:50for patients with higher tumor burdens
• 09:52or are in neoadjuvant settings,
• 09:54or in cases where resection or
• 09:55ablation is in a location considered
• 09:58terribly difficult or dangerous.
• 10:00He We see a case of a patient with HIV,
• 10:03cirrhosis and a 4.8 centimeters segment.
• 10:068 HTC Embolization was performed
• 10:07to maintain the patient on the
• 10:10transplant waiting list.
• 10:11We see completing the Krosis,
• 10:13and she alternately underwent a transplant.
• 10:15This is a patient with compensated HTP,
• 10:18sarot sis, Anna Sala,
• 10:20Terry HTC and segment 8 and a platelet
• 10:23count of 57 who also needs treatment.
• 10:25There's a bridge to transplant.
• 10:27This tumor is shown here is in
• 10:29a very difficult location,
• 10:31and she successfully underwent
• 10:33embolization with no residual tumor,
• 10:35and this patient also underwent
• 10:36liver transplantation.
• 10:40Here we're able to see a patient that was
• 10:43treated in their 90s who has a large HTC.
• 10:45In this case, the ACC actually
• 10:47ruptured through the capsule,
• 10:48and as you probably are all aware,
• 10:50this is associated with poor, if not dismal,
• 10:53prognosis were able to successfully
• 10:54treat it with radio embolization,
• 10:56which was done as an outpatient,
• 10:58and this can be seen here.
• 10:59She had complete response and lived
• 11:01up to five after four years later,
• 11:03and then Lastly.
• 11:04This is a case of a patient that has BI lo
• 11:07or HTC with extensive portal vein tumor.
• 11:10Rambus, so typically this patient would
• 11:12have less than six months to live,
• 11:14but as expected, this is a relatively
• 11:17young patient with the family,
• 11:19small young children, etc.
• 11:20So we try to offer procedure to help this
• 11:23patient an we did Chemoembolization,
• 11:25and Fortunately we were successful in
• 11:28prolonging his life for about 16 months.
• 11:32So to wrap up,
• 11:32I just want to give you my five year of
• 11:35vision for a modern academic evidence
• 11:37based intervention on college program.
• 11:39First we want to increase the IO
• 11:41clinical research programs throughout
• 11:42the smilow cancer hospital and
• 11:44kiss centers throughout the area.
• 11:46Further,
• 11:46it will be important to improve the
• 11:48education for the community provider of
• 11:50the procedures that we offer and which
• 11:52patients are more likely to benefit.
• 11:54We also need better integration of care
• 11:56into current and future oncology workflows.
• 11:58We will be working on initiating some new
• 12:01clinical trials related to liver oncology.
• 12:03As discussed last week
• 12:04on liver generation was,
• 12:06which was dragging one study,
• 12:07but we also are going to be
• 12:09involved in the leap 012 study,
• 12:11which is going to be an immunotherapy
• 12:14chemo embolization plus or minus
• 12:15immunotherapy that we're going to be
• 12:17initiating or starting very soon.
• 12:19And then it's also going to be
• 12:21critical to have faculty who are
• 12:23sub specialized to be able to
• 12:25provide this level of advanced air.
• 12:28We also plan to make image guided
• 12:30biopsies more readily available and
• 12:32hoping to improve turn around time.
• 12:34There have been ongoing discussions
• 12:36with Doctor Chen Lu as you all
• 12:39know is the chair of pathology.
• 12:41To work on a by an image guided
• 12:44by over Pozza Tori.
• 12:46As will be seen from the next speakers,
• 12:48we are currently working on advanced
• 12:51preclinical and Translational
• 12:52research with a focus on molecular
• 12:54imaging immuno oncology.
• 12:54We're also very active in machine
• 12:57learning and artificial intelligence as
• 12:59it relates to liver cancer and based
• 13:01on these interesting topics we hope
• 13:03to be able to achieve and expand our funding.
• 13:05And Lastly we have a newly
• 13:07established IR residency program.
• 13:09In this we would like to develop an
• 13:12Interventional Oncology Fellowship and
• 13:13we are looking into the possibility of
• 13:16applying for a T32 training grants.
• 13:18So with that,
• 13:19these are visions and goals I'd like
• 13:22to then introduce my Co director
• 13:24for the Interventional Oncology
• 13:26Research Lab at Yale,
• 13:28who will be discussing
• 13:29quantitative biomarkers,
• 13:30molecular imaging in artificial intelligence
• 13:32to guide the therapy of liver cancer.
• 13:35So thank you for your attention.
• 13:46Alright, thank you very much everyone.
• 13:48I'm really excited to be presenting this
• 13:50topic on behalf of our lab and this is
• 13:53more of a vision presentation of what we've
• 13:55accomplished and a couple of thoughts
• 13:57that we put together over the last years.
• 13:59None of this would have been possible without
• 14:01really vast infrastructure of collaborators,
• 14:03which I'll be talking about
• 14:05a little bit later.
• 14:06I want to say that I very excited about.
• 14:09David joining the group here and really being
• 14:11our new leader and Interventional Ecology,
• 14:14and we all completely subscribed to
• 14:16the vision that she just provided.
• 14:18So as you know,
• 14:20I'm going to be focusing on the liver cancer,
• 14:23and primarily I'm going to start
• 14:25off with a BC else staging system.
• 14:28The most recent one was released in 2018,
• 14:30and even in that more recent addition to
• 14:33it we have seen a clear separation between
• 14:36intermediate station advanced stage disease,
• 14:38so they exist.
• 14:39Apparently in two separate planets,
• 14:41however,
• 14:41really in clinical practice and in signs,
• 14:44this is very transitional area where both
• 14:46seemed to be interacting quite extensively.
• 14:48We saw a lot of clinical trials
• 14:51focusing an overlap of those therapies,
• 14:53and this is exactly the realm where
• 14:55most patients are being diagnosed
• 14:57and also the realm where you know
• 14:59we think there is most room of
• 15:02improvement for patient outcomes,
• 15:04so systemic therapies for HTC then
• 15:06and now I borrowed the slides for
• 15:08from Agusan Abou Alfa, who's?
• 15:10Medical Oncologist at Memorial Sloan
• 15:12and this is how the market look like.
• 15:15Five years ago we just had one drug
• 15:17and this it's not even a complete view.
• 15:20This is where we are right now.
• 15:22So with him just five years,
• 15:24we have so many drugs being approved and
• 15:26most of 'em actually are very dedicated,
• 15:29specific targeted molecules
• 15:30and checkpoint inhibitors.
• 15:31So this flood of novel agents
• 15:33actually probably means the left
• 15:34ship of systemic therapy in the CL.
• 15:36See staging system and what we will
• 15:38witness is a greater overlap between local,
• 15:41regional and.
• 15:42Systemic therapies of immunotherapy.
• 15:44And this is where we come from.
• 15:47We focused now on the biomolecular mechanisms
• 15:49in the tumor micro environment behind
• 15:51those therapies and their combination.
• 15:53So we focus on two of the well known
• 15:56hallmarks of cancer that are particularly
• 15:58relevant for HTC specifically on its
• 16:00ability to avoid immune destruction
• 16:02and the deregulation of seller
• 16:04energetic into metabolic phenotype.
• 16:05So let's have a look at the HTC cell,
• 16:09so we now it has a very pronounced
• 16:11Warburg effect,
• 16:12and this is specifically a cell
• 16:14that is extremely dependent on
• 16:16the glucose metabolism.
• 16:17And so called the Arabic and
• 16:19anaerobic glycolysis, and well,
• 16:21the glucose is being taken up with that sell.
• 16:24It produces lactate.
• 16:25It is that essentially is
• 16:27D Cup from the Krebs cycle,
• 16:29and this lacked.
• 16:30It is then being flooded directly
• 16:32into the extra seller micro environment.
• 16:34And because of that there was
• 16:36dramatic acid ification of the
• 16:38surrounding tumor micro environment.
• 16:39And what does that really mean for us?
• 16:42The extracellular pH and liver cancer,
• 16:44especially aggressive age species
• 16:45known to be very, very low.
• 16:47And now finding say that this is a very
• 16:50important exclusion mechanism for four.
• 16:52Immune cells and the mechanism
• 16:54of immune cell exhaustion.
• 16:56So this has been previously reported
• 16:58and we now know more than ever that
• 17:01actually tumor infiltrating T lymphocytes
• 17:03depend extremely on extra seller acidity
• 17:06and competitive glucose deprivation,
• 17:07and that the reversal of this hostile
• 17:10tumor micro environment can actually
• 17:12improve their infiltration into the tumor.
• 17:14An immune response.
• 17:16So to sum it up,
• 17:18loads for seller pH is a marker of
• 17:21tumor aggressiveness and the presence
• 17:22of Protons and lactate really protects
• 17:25cancer against an immune response.
• 17:27It promotes new angiogenesis.
• 17:28The lactic acidosis plays in multifaceted
• 17:31role that is not fully understood for
• 17:33all variable immune cell subgroups,
• 17:35and the high lactic levels are
• 17:37actually strong predictors of
• 17:38metastatic disease and poor outcome.
• 17:40So where does all this fit in with taste?
• 17:43So theoretically this is an embolus therapy
• 17:46that changes the tumor micro environment.
• 17:49And what happens is that in theory
• 17:51tastes me potentially exacerbate
• 17:52a hostile tumor micro environment
• 17:54that attenuates treatment efficacy.
• 17:55Let's look at it.
• 17:57I mean,
• 17:58it induces a hypoxic injury that potentially
• 18:00may exacerbate the low pH even further.
• 18:02It potentially stimulates brand
• 18:04eugenic signaling and the embolization
• 18:06even might prevent delected
• 18:07transport away from the tumor.
• 18:08So postdates tumor marker environment
• 18:10was thought to be immune inhibitory,
• 18:12but is that really true?
• 18:15I mean,
• 18:15let's look at HTC in general,
• 18:17where it stands in terms of Genomic
• 18:20somatic mutations as opposed
• 18:21to Melanoma and lung cancer.
• 18:23Really, HTC is somewhere in the middle,
• 18:25so the idea of the Immuno and inoculation
• 18:28with locoregional therapies with
• 18:29initially discarded because of that.
• 18:31So,
• 18:32however,
• 18:32let's have a look at the pathology
• 18:34of a tumor treated with taste.
• 18:36This is a patient with HTC.
• 18:39Histology was taken 11 days post locoregional
• 18:41therapies that was transplantation.
• 18:42What we see we see in necrotic
• 18:45tumor packed with in this
• 18:46particular case drug eluting beads.
• 18:48And surrounding that we see a very
• 18:51dense immune infiltrate and if
• 18:52we zoom in We really almost see
• 18:53a secondary immune follicles and
• 18:55macrophages and T cells really
• 18:57accumulating the transitional zone.
• 18:58But they can't really penetrate
• 19:00into the necrotic areas of the tumor
• 19:02where all the good end to Jenison.
• 19:04Why is that?
• 19:04So our mission was at this point
• 19:07to really look at the underlying
• 19:09mechanisms and how do systemic therapies.
• 19:11How does in the immune system and
• 19:14local regional therapies interact?
• 19:15So we wanted to develop imaging
• 19:17instruments for noninvasive functional
• 19:19characterization monitoring of the
• 19:20tumor micro environment in the setting
• 19:22of the immune response and taste,
• 19:24and that was the mission of the
• 19:26lab that we essentially star
• 19:27started several years ago.
• 19:29So the one of the first projects that
• 19:31I'm going to talk about is recently
• 19:34published paper clinical Cancer Research.
• 19:36That focused on the development of an
• 19:38extracellular pH probe that demonstrates
• 19:40extracellular pH noninvasively
• 19:42using spectroscopic methodology,
• 19:43and I'm actually really appreciate it.
• 19:46'cause that partnered in design and
• 19:48work on the on this novel mechanism
• 19:51with Daniel Komen and if I meet
• 19:54higher from the Mr Research Center
• 19:56who have been really instrumental
• 19:58and great partners in developing.
• 20:01And we used to model of a rabbit liver tumor,
• 20:04which is so far the only moderate size animal
• 20:07model that has a faithful reproduction.
• 20:09Reproduction of an HTC tumor
• 20:11environment and the has a high
• 20:13clinical relevance because we can
• 20:14do the embolization also image those
• 20:16animals in real size clinical scanners.
• 20:19We do have a 3T MRI system for that
• 20:21and also cutting edge IR suite which
• 20:24we use in the white Rick with the help
• 20:27and support of Alston is this group.
• 20:29So this is an immuno competent host.
• 20:32Anna hyper glycolytic metabolic phenotype.
• 20:34An essentially those tumors have
• 20:35been characterized previously,
• 20:37is very reminiscent of the actual
• 20:39tumor marker environment of HTC.
• 20:40So what is a good way to image pH so
• 20:43their sensor jeans that you can use
• 20:46their luminescent probes that you can
• 20:48translate yourselves with reporters?
• 20:50But what we in radiology aim for is
• 20:52really the noninvasive measurement
• 20:54affects your seller pH and this is
• 20:56where our msit based spectroscopic
• 20:58birds methodology comes into place.
• 21:01This has been established previously
• 21:02in brain tumors by Daniel.
• 21:04And we have translated this
• 21:06essentially to deliver.
• 21:07This is NMR Spectra,
• 21:08scopic method that measures the
• 21:10redundant deviation of shifts and
• 21:12temperature in the region of the tumor,
• 21:14and we can generate the 3D extracellular
• 21:16pH map that provides us with essentially
• 21:18a very accurate characterization
• 21:19of the tumor micro environment.
• 21:21So if we look at the tumor in by
• 21:24itself and measure it as compared
• 21:26to the surrounding liver tissue,
• 21:28we see that the tumor in by itself
• 21:30has much lower pH and baseline in an
• 21:33untreated fashion and the surrounding liver,
• 21:35and that's important.
• 21:36And we know that this is probably
• 21:38due to the overt Warburg effect and
• 21:41overexpression of glued one and lamb two
• 21:43as indicators for the micro environment.
• 21:45Now, what does taste actually do with it?
• 21:48And this is where the surprising
• 21:50result from our study came in.
• 21:51We looked at the effects of
• 21:53embolization with little one day,
• 21:55one week,
• 21:56and two weeks after we actually
• 21:58treat those tumors and measured the
• 21:59pH and as opposed to most of the
• 22:02assumptions that taste is going to
• 22:04actually exacerbate the The Anti.
• 22:06Yeah, I mean the logic, the immune,
• 22:09evasive tumor micro environment,
• 22:11it actually did the opposite.
• 22:13So what we see is a normalization
• 22:15of the tumor pH towards almost
• 22:18delivered background levels.
• 22:19Overtime after taste.
• 22:20And that is an important finding
• 22:22because that gives us an opportunity
• 22:25to use local regional therapy in
• 22:27preparation for successful anti
• 22:29checkpoint checkpoint inhibition
• 22:30therapy and we actually demonstrated
• 22:32this also with a more direct and
• 22:35exacerbated model. Where we.
• 22:37Eustace and actually injected bicarbonate
• 22:39directly with taste and what we
• 22:41demonstrated here is that on baseline
• 22:43you have an acidic tumor with taste.
• 22:45You have a minor improvement of pH towards
• 22:48normal over overnight essentially,
• 22:49but if you add bicarb you can
• 22:52immediately improve the pH almost
• 22:53essentially to normal and yellow and
• 22:55green means higher levels than blue.
• 22:58If we now look at, for example,
• 23:00H Lady are expression at baseline,
• 23:02we see immune cell exclusion were really see
• 23:05all the immune cells in the tumor periphery.
• 23:08And with with taste alone,
• 23:09immediately at least one day after therapy,
• 23:12you don't really see a
• 23:13lot of immune invasion,
• 23:14but with taste and bicarb,
• 23:16essentially improving and elevating a pH,
• 23:18you see massive intratumoral
• 23:19infiltration of immune cells,
• 23:21and that's an important finding that
• 23:22we made here in actually an important
• 23:25other thought that we should be thinking
• 23:27about as a group generally in oncology,
• 23:29and what kind of agents we use.
• 23:32So when we use a key mobilization with oil,
• 23:34which is Lupito, we achieved,
• 23:36achieves seemed to be achieving
• 23:38very different effects.
• 23:39Compared to,
• 23:39for example,
• 23:40using an cuisine beads or Lumi beads,
• 23:42which is another flavor of beads
• 23:44and that is important for us to
• 23:46understand because through the
• 23:48uncle logic community taste was
• 23:49taste for long period of time,
• 23:51but in reality it is not because we
• 23:53know that different embolic particles
• 23:54will induce very different in a logical
• 23:57effects in very different changes
• 23:59to the tumor marker environment.
• 24:00And that is something that we
• 24:02really need to study,
• 24:04especially if we want to combine the
• 24:06local regional therapies with systemic
• 24:07therapies and checkpoint inhibition.
• 24:09Now another level of investigation
• 24:11that we took to token published
• 24:14recently in radiology,
• 24:15again close collaboration with
• 24:17Daniel Komen Fahmideh Hyder but
• 24:20also other groups like rip Ocala's
• 24:22group and also want to mention Ruth,
• 24:25Montgomery,
• 24:25and Joshi,
• 24:26who helped us generate this research.
• 24:29Here we wanted to demonstrate
• 24:31that we can actually use molecular
• 24:33dedicated inmar probes nanoprobes
• 24:35labeled antibodies to visualize
• 24:37the immune system surrounding those
• 24:39tumors in vivo and noninvasively.
• 24:42So what we did is we use both iron
• 24:44oxide particles and the direct
• 24:46injection of gadolinium labeled
• 24:48antibodies to visualize those tumors
• 24:50and put those animals with those
• 24:53implanted tumors into them are what
• 24:55we demonstrated was that we were
• 24:57able to clearly delineate microfusion
• 24:59immune cell infiltration and macrophage
• 25:01and immune cell accumulation in
• 25:03the periphery of those tumors.
• 25:05On T2 MRI sequences,
• 25:07and that was proven in histologically
• 25:09with Prussian blue staining,
• 25:11which stands for iron.
• 25:12And we saw that the spines,
• 25:15the iron oxide particles really
• 25:17deposited and CD11B positive macrophages
• 25:19and we confirmed that later with
• 25:21CD 68 staining and the same thing
• 25:24is true for our induction injection
• 25:26of gadolinium labeled antibodies.
• 25:27You probably heard about this from
• 25:29the molecular imaging using pet,
• 25:31but here we have a much Marsh
• 25:33higher focal resolution where we
• 25:35label antibodies with gadolinium
• 25:37and then inject them
• 25:38directly into those tumors and through
• 25:41the artery, and we can see that there is
• 25:44very specific accumulation and staining.
• 25:46Of immune cells and perforate the tumor
• 25:48that we can now really locali characterize,
• 25:51and that's that's a big step forward
• 25:53because not only can we now image the tumor
• 25:56marker environment from a pH perspective,
• 25:58we can also images from a presence
• 26:01and functionality of immune cells.
• 26:03So for the first part of the conclusions
• 26:05are their appearance of, for example,
• 26:07more than five targeted agents to
• 26:09treat HTC will cause a left shift
• 26:12of systemic therapy in the BC else,
• 26:14and we really must do some heavy lifting.
• 26:16It's up to art actually to put
• 26:18to put the collaboration between
• 26:19the Interventional in college and
• 26:21Immuno Oncology on the right track.
• 26:23From a science perspective we must
• 26:25also work more on non invasive
• 26:27imaging modalities and we did so with
• 26:29the extracellular pH that revealed
• 26:31that ace actually is an inducer.
• 26:33Profound changes to tumor marker
• 26:35environment that may help immunotherapy
• 26:37to be more effective,
• 26:38and pH is key for us as marker,
• 26:42but the changes in pH depends on
• 26:44the embolic materials that we use.
• 26:47We also now know that C test would
• 26:49lipiodol may achieve a partial reversal
• 26:52of the immune evasive tumor marker
• 26:54environment on itself two weeks after taste,
• 26:57and that we can use them are
• 27:00instruments to actually detect immune
• 27:02cells directly surrounding the tumor.
• 27:04So in summary,
• 27:05the for the first part of my conclusion
• 27:07is that local regional therapy
• 27:09and intervention alone college.
• 27:10You must be established in combination
• 27:12with each other.
• 27:13We have to have a synergy between
• 27:15the two iOS interventional and Immuno
• 27:17Oncology and it's up to us to work on that.
• 27:20And that generates another bigger
• 27:21problem that we really workup on very
• 27:23closely with Jim Duncan and collaboration
• 27:25with that group and biomedical engineering.
• 27:27And this is the overwhelming growth
• 27:29of data and will be talking about
• 27:31this little bit later.
• 27:32Tall is going to be presenting on that,
• 27:35so we know that the data that we
• 27:37have specifically in imaging and.
• 27:39Cancer in general explode so we
• 27:41have more than 20 times more data
• 27:43today than we have in 2013,
• 27:45and so that is a huge challenge.
• 27:47Anne, it's David already presented.
• 27:49We have different flavors
• 27:51of chemo embolization,
• 27:52embolization with drug alluding
• 27:53Beatson Radio mobilization.
• 27:54We have a variety of different
• 27:56methodologies in ways how we
• 27:58can introduce tumor necrosis,
• 28:00cold heat,
• 28:00and electrocution in a way we have
• 28:03five different and more coming
• 28:05up with systemic therapy agents,
• 28:07all of which pose different changes
• 28:09to the tumor micro environment.
• 28:11So we went overwhelming burden of
• 28:13complex data that actually may
• 28:14impede effective clinical practice.
• 28:16We need to address that.
• 28:18And so,
• 28:19how do we transform the data
• 28:21from burden to value?
• 28:22We know that we have an increase in data.
• 28:25We have very complex medical health records.
• 28:27We have genomics sequencing information.
• 28:29We have the availability of complex
• 28:31computer based algorithms and
• 28:32now the availability, availability of
• 28:34computational power at a lower cost.
• 28:36And that is the space where
• 28:38artificial intelligence now comes in.
• 28:39And this is essentially data driven.
• 28:41Learning that deep down can be
• 28:43explained as machine learning that
• 28:45recognizes trends and objects in pre
• 28:46labeled patterns or deep learning,
• 28:48and tall is going to be talking
• 28:50more about that where?
• 28:52We use those networks to actually
• 28:54learn from data without pre labeling
• 28:56the outcome and that may hopefully
• 28:58increase the workflow efficiency,
• 29:00improve our diagnostic accuracy really
• 29:02enabled predictive recommendations
• 29:03for us Taylor Personalized
• 29:04Therapeutic recommendations and help
• 29:06us introduce Precision Medicine,
• 29:07but caution is still very important and
• 29:10AI is being hyped and I think we need
• 29:13to approach it in a gradual fashion.
• 29:16Really see what is doable and what's not.
• 29:19So how can this advanced data
• 29:22analysis help us?
• 29:23It can improve the diagnosis from automation
• 29:25and introduction of novel biomarkers.
• 29:27It can also help us make therapeutic
• 29:30decisions and probably introduce a
• 29:32level of better personalized care.
• 29:34Hopefully we will be able to ultimately
• 29:36improve Inter procedural guidance,
• 29:38especially with now with the
• 29:40introduction of robotics into IR,
• 29:41and specifically follow up imaging.
• 29:43We hope that in the realm of tumor
• 29:46response and patient patient outcome
• 29:48prediction directly after the therapies,
• 29:50those technologies will help
• 29:52us put all this data together.
• 29:54Now,
• 29:55prior to introducing the other
• 29:56speakers and talking and giving
• 29:59specific abstract presentations.
• 30:00The two larger topics that I presented.
• 30:03I want to really thank the sources of
• 30:05funding inspiration mentoring in our
• 30:07community. Already mentioned Drake.
• 30:08I wanted to thank Jim sincerely and
• 30:11Todd and and David who joined the lab.
• 30:13Also great partners of mine are
• 30:15Ruth and Nick and I really,
• 30:17really appreciate those collaborations
• 30:18that are increasingly interdisciplinary,
• 30:19but at the same time I wanted to
• 30:21also point out that mean radiology
• 30:23have a very large biomedical
• 30:24imaging collaborative network,
• 30:26and I want to thank from Eden, Daniel,
• 30:29and and Larry and everyone mentioned here.
• 30:31I'll say I'll.
• 30:32From White Rick,
• 30:33who have been instrumental over the last
• 30:34five years in my personal development
• 30:36and also in the development of the lab,
• 30:38and I think that is the environment
• 30:39that we need and we need to highlight
• 30:42that to the Cancer Center and
• 30:43introduce it to to all of you.
• 30:45And I'm very happy to do so here.
• 30:47And,
• 30:47you know,
• 30:47at this point I would like to thank on behalf
• 30:50of the Interventional College lab which
• 30:51I'm privileged to Co direct with David and
• 30:53all of our sponsors and really say that,
• 30:55you know,
• 30:56there's one thing across all this.
• 30:57There is a single light of science,
• 30:59then to brighten it anyways,
• 31:00to brighten it everywhere.
• 31:01This is our group model by Isaac Asimov.
• 31:03And with that,
• 31:05I'd like to introduce two or
• 31:07students that work with us.
• 31:09First,
• 31:10Jessica Santana.
• 31:10She's going to be talking about
• 31:13taking the molecular imaging
• 31:15really to next level in our new
• 31:17animal models, and tall is going to be
• 31:20talking about his role as Biomedical
• 31:23Engineering graduate student within
• 31:25the bio medical imaging Sciences on
• 31:28focusing on predicting outcome and
• 31:30recurrence of HTC after intervention.
• 31:32And I'd like to.
• 31:34And now the stop sharing the
• 31:35screen and give it to Jessica.
• 31:46Hi all, I'm Jessica.
• 31:47I am a master level demonologist
• 31:50Anna graduate student at the
• 31:52Yale Interventional College Lab
• 31:54and I'm very excited today to
• 31:56be presenting my work, titled,
• 31:58Noninvasive molecular imaging
• 31:59allows characterization of the
• 32:01immune response following hepatic
• 32:03radiofrequency Ablation in a mouse model.
• 32:07Nothing to disclose,
• 32:08so giving you a brief overview
• 32:11on hepatocellular carcinoma.
• 32:12So as we know, this is a classical
• 32:15inflammation associated Carcinoma and
• 32:16now is the third most common cause
• 32:18of cancer related death worldwide.
• 32:21With the majority of patients being
• 32:23treated with locoregional therapy as
• 32:25they may alternative option over surgery.
• 32:27However, the problem with locoregional
• 32:29therapy is that a significant fraction
• 32:32of patients they tend to recur and the
• 32:35causes for recurrence can vary a lot.
• 32:37And it has been suggested, for example,
• 32:40that the immune response to radiofrequency
• 32:43ablation can play both both roles.
• 32:45An Protogenic Side effects as well as
• 32:48abscopal effects that in turn can positively
• 32:51impact the immune response to cancer.
• 32:54However,
• 32:54we have currently no instruments that
• 32:57would allows us to non invasively
• 33:00monitor such immune response.
• 33:02So the purpose of this work is to
• 33:04develop a noninvasive molecular imaging
• 33:07instruments to visualize such immune
• 33:09response to thermal injury following RFA.
• 33:14So our group has built a translation
• 33:17of mouse model of radiofrequency
• 33:19ablation as a platform to develop and
• 33:22validate mirbase dimona probes for in
• 33:24vivo imaging of the immune system.
• 33:27And based on our findings,
• 33:29we have observed that there is a after
• 33:33radiofrequency ablation and normal liver.
• 33:35There is a strong time dependent local
• 33:38infiltration of immune cells that
• 33:41orchestrates the tissue healing process
• 33:43and there is an overtime transitional
• 33:46zone of those immune cell infiltrate.
• 33:49So having a specific cell population
• 33:51locally present at the transitional
• 33:53zone between the necrotic area and
• 33:56the normal liver parenchyma at a
• 33:58specific time point serves us as
• 34:00an illegal platform to build and
• 34:03validate our dedicated immune probes.
• 34:06So guiding you through our
• 34:08experimental design.
• 34:09We have a bladed and normal liver of a mouse,
• 34:13and after characterizing a large
• 34:15infiltration of a specific cell
• 34:17population is specific time point.
• 34:19We have established a dedicated
• 34:21gadolinium labeled antibody that was
• 34:24delivered systemically to target
• 34:26a specific cell population at the
• 34:28chosen time Point Post Ablation.
• 34:31And a second imuna probe also used
• 34:33in this study,
• 34:34is the small iron oxide particles,
• 34:37and we know that RN particles they
• 34:39have been largely used in the clinical
• 34:41setting as the dark contrast agent and
• 34:44once they are delivered systemically,
• 34:46they are able to be phagocytosed by
• 34:49circulating phagocytes that once they
• 34:51migrate to the site of inflammation
• 34:54they can cause a local deposition
• 34:56of Iran and this study we have
• 34:58demonstrated that both gadolinium
• 35:00labeled antibodies inspire probes.
• 35:01They were able to be imaged using a
• 35:05higher resolution Mr animal scanner.
• 35:08So in our first setting of experiments
• 35:11we have demonstrated that the
• 35:13ablation zone itself can be easily
• 35:15imaged on a 9.4 Tesla Burger.
• 35:17So here on your left you have
• 35:20the pictures taken seconds after
• 35:22ablation to show you or give you an
• 35:25idea of how the ablation site looks
• 35:27like an one week post Ablation,
• 35:30and we have delivered pure gadolinium
• 35:32systemically and we ran a T1
• 35:34weighted MRI sequence and we could
• 35:36see precisely the Ablation.
• 35:38Side and on your right we have an ex
• 35:40vivo confirmation of what we see on them,
• 35:42right?
• 35:44But you might have been wondering
• 35:46why exactly 7 days post ablation.
• 35:48So to the best of our knowledge,
• 35:51we know that our our thing doost
• 35:53thermal tissue injury,
• 35:54largely contributes to a strong time
• 35:57dependent innate immune response at
• 35:59the margins of the necrotic zone.
• 36:01And according to our findings,
• 36:03we have observed the largest
• 36:05accumulation of city 68 positive
• 36:07macrophages in the transitional zone,
• 36:09precisely seven days post ablation.
• 36:11So this is the time point we decided
• 36:15to base our experiments on.
• 36:18So we have decided to deliver systemically.
• 36:21Those aren't oxide particles.
• 36:23Seven days post Ablation and we
• 36:26have confirmed that there is a local
• 36:29deposition of those iron oxide
• 36:32particles exactly 24 hours after
• 36:34systemic delivery. At the transitional zone.
• 36:37So here we have an ex Vivo Prussian
• 36:39blue staining,
• 36:40although one week post ablation,
• 36:42mouse liver 24 hours after
• 36:44systemic delivery of Spions and
• 36:46when we image those animals,
• 36:48we could demonstrate any Viva local
• 36:50deposition of those phagocytes
• 36:51at the transitional zone.
• 36:53So here on your left you have a teacher
• 36:56weighted MRI of a one week post Ablation,
• 36:5924 hours after systemic delivery
• 37:01of Spions and as a Redpath
• 37:03correlation we confirm the ex vivo
• 37:05specifically position of the spine.
• 37:07At the transitional zone,
• 37:093 min inflorescence and
• 37:10Prussian blue staining.
• 37:14And As for our gadolinium based
• 37:17immuno probes, we have used a anti CD
• 37:1968 antibody tagged with gadolinium.
• 37:22So after XP will observing a massive
• 37:24infiltration of city 60 positive
• 37:26macrophages in the prohibition of
• 37:28zone we delivered gadolinium tagged
• 37:30with city 68 systemically and was
• 37:33able to see a specific position of
• 37:36those cells in the Prohibition of rim.
• 37:38So here on your left you have the Ablation.
• 37:43Cite the picture taking seconds after
• 37:46ablation showing the ablation site.
• 37:48And we also have 81 way to MRI when
• 37:51we post Ablation After 24 hours with
• 37:55gadolinium city 68 delivered systemically.
• 37:59And on your left you have the picture taken,
• 38:02ex Vivo confirming of what we're
• 38:05seeing on the MRI.
• 38:07And to confirm that this parable
• 38:10itional darkroom is seen precisely
• 38:12in animals receiving those city,
• 38:1568 tagged to 68 antibodies tagged
• 38:18with gadolinium.
• 38:19We also ran T1 weighted MRI with
• 38:23pure gadolinium.
• 38:24So on your lap to have the baseline
• 38:27and it comparison,
• 38:29we have a T1 weighted MRI with pure
• 38:33gallium injected and we have confirmed
• 38:35that only with animals receiving city
• 38:3868 tagged with get alignment we have
• 38:42this precise parable itional rim
• 38:44showing a local infiltration of CD
• 38:4716 positive macrophages and therefore
• 38:49in vivo visualization of those cells.
• 38:53We have also confirmed the ex vivo
• 38:55a specific labeling of immune cells
• 38:58using imaging mass atama tree.
• 39:00So he ran on your left.
• 39:02We have the T1 weighted MRI of one
• 39:04week post ablation after gadolinium
• 39:06labeled antibody administration,
• 39:08where you can see the rim of the
• 39:11local deposition of the infiltrating
• 39:13cells an on your right.
• 39:15We have the X visual confirmation with
• 39:17the image Ng Masama tree of local
• 39:20deposition of the sea to 68 macrophages.
• 39:23In the transitional zone.
• 39:26So as the main conclusions
• 39:28and findings of this study,
• 39:29it tells us that both spines and
• 39:31Catalina based molecular imaging
• 39:33allows for specific labeling
• 39:34of local immune infiltrate,
• 39:36and this is also a translation of
• 39:38study with the proof of principle
• 39:40for the visibility of the MRI imaging
• 39:42for of macrophages on a 9 point.
• 39:45For Tesla MRI scanners.
• 39:46And also tells us that noninvasive
• 39:49in vivo detection of the immune
• 39:51system can be achieved using
• 39:53dedicated immune probes.
• 39:55An ask for our future perspective
• 39:59and clinical application.
• 40:00We can have this as a useful tool to
• 40:03study and characterize the interplay
• 40:06between the tumor micro environment
• 40:08and the cell opinion selectivity in
• 40:11vivo and also gives us the possibility
• 40:13to integrate complimentary molecular
• 40:15MRI imaging of the immune system.
• 40:17And let's say the extrasolar pH
• 40:19liver cancer model for simultaneous
• 40:22characterization of this immuno
• 40:24metabolic cross dock.
• 40:25This also serves us as a platform to
• 40:28study strategies for local modulation
• 40:30of the immune microenvironment
• 40:32towards a moon are permissive
• 40:34phenotype and can also serves us as
• 40:37if they were gnostic immunotherapy.
• 40:40So I'd like to thank you for the
• 40:42opportunity to present this work
• 40:43as well as the yield bio medical
• 40:46imaging Department. And now I'll give
• 40:48you the word to my friend, tall.
• 40:58Hello everyone, my name is Charles
• 41:01early and I'm graduate student in
• 41:03the Biomedical Engineering program
• 41:05here at Yale and I'm a member of the
• 41:09Interventional Oncology lab since 2018
• 41:11and today I'm going to be presenting
• 41:13our project on deep learning use to
• 41:16predict disease recurrence of HTC,
• 41:18based on NMR imaging.
• 41:23So a little bit about ATC,
• 41:25so HTC is the primary tumor of delivered.
• 41:28It usually develops in the
• 41:30setting of chronic liver disease,
• 41:31and while the diagnosis of ATC
• 41:33could be made by imaging alone,
• 41:36sometimes a biopsy may be required
• 41:38to support this diagnosis.
• 41:40A little bit statistics every 40 seconds at
• 41:44patient diagnosed with ATC in 2020 alone,
• 41:48the death rate is approximately
• 41:51the 800,000 deaths worldwide.
• 41:53ATC may recur and there is no
• 41:56significant imaging biomarker or
• 41:59clinical biomarkers to reliably predict
• 42:02recurrence before location to treatment.
• 42:06One of the treatment is
• 42:08liver transplantation.
• 42:08It helps to decrease the
• 42:10chance of disease recurrence.
• 42:12However,
• 42:12we are all aware to the shortness of organs.
• 42:16Therefore, 2 criterias of
• 42:17allocations of levers are being used.
• 42:19One of them is the Milan criteria,
• 42:22which was presented in 1996 and
• 42:24the other one was presented by
• 42:26the University of California,
• 42:28San Francisco on 2001 basically
• 42:30extended this Milan criteria.
• 42:32Both of them are based on low
• 42:34level handcrafted features such as
• 42:36tumor size and number of tumors.
• 42:39Ann even dog.
• 42:40Using these criteria,
• 42:42we are seeing 15 to 20% of transplanted
• 42:45patient to occur within the 1st five years.
• 42:48So as I said,
• 42:50these two criterias location criterias
• 42:52has suffered from false positives,
• 42:54and our hypothesis is that there is more
• 42:57information in radiological images,
• 42:59specifically MRI that correlate
• 43:01with HTC recurrence.
• 43:02Then the naked eye could detect
• 43:04and the way we're going to try to
• 43:08test this hypothesis is by using
• 43:10deep learning algorithm to extract
• 43:12features from MRI images and try to
• 43:16use them to correlate to ATC re occurrence.
• 43:20And a little bit before I start
• 43:22talking about our methodology,
• 43:24I will talk about about data
• 43:26driven predictive modeling and
• 43:27a little bit of deep learning.
• 43:29So when we talk about data
• 43:31driven predictive model,
• 43:32we usually refer to two different variables,
• 43:34the Explanatory variable and explain
• 43:36part of our target and we want to
• 43:39take these explanatory variables
• 43:40to feed them into predicted model
• 43:42which will give up the outcome of
• 43:44our target building which could
• 43:46be recurrent or not.
• 43:47And this predictive models
• 43:49has different shapes.
• 43:50And algorithms which you may hurt.
• 43:52For example your networks this season.
• 43:55Trees as VM's.
• 43:56All in all,
• 43:57these algorithms that try to do the
• 43:59same thing they try to create or to
• 44:02estimate a mathematical function that
• 44:05correlated this input into doubt.
• 44:07And there are two main elements
• 44:09of this estimation process.
• 44:11One of them were trying to see
• 44:12what is the mathematical element
• 44:14interaction between our features
• 44:16and the mathematical operations.
• 44:18That's going to be incorporated
• 44:20within these mathematical function.
• 44:21For example,
• 44:22the multiplication of X1 and X2
• 44:24and the second thing we're trying
• 44:26to estimate the weight of these
• 44:28mathematical elements within this function,
• 44:31which could be seen here as better 0,
• 44:33better one, etc.
• 44:36And I'll give a short and very
• 44:38simplistic example of these
• 44:39data driven predicted model.
• 44:41So here we are,
• 44:42having having only one variable Explanatory
• 44:44Variable, which is the age and we are
• 44:47trying to predict the disease patient
• 44:49is positive or negative to the disease.
• 44:52So for example in the right
• 44:53side of the screen you can see
• 44:56this mathematical function.
• 44:57Why would be considered positive if the age
• 45:00is greater than better otherwise negative?
• 45:02And as we get more and
• 45:04more examples we could.
• 45:06Estimate this better.
• 45:07For example,
• 45:08here we are estimating better to be 60,
• 45:12but as more data comes we
• 45:15can update modifier.
• 45:17Wait and basically change our function.
• 45:21Ann everything is OK till we have
• 45:23a data example that prevent us from
• 45:25creating one to use this mathematical
• 45:27function to a separate between
• 45:29between these these two groups.
• 45:31So here we need to incorporate a new feature,
• 45:34for example wait when we are doing that.
• 45:36We can create a more complex function to
• 45:39separate these two groups and basically
• 45:41this is what we're going to try to do.
• 45:43Today we're going to try to find the
• 45:46features that will allow us to separate
• 45:49between the recurrence in Nonrecurring.
• 45:51When we're talking about deep
• 45:53learning where basically usually
• 45:55refer to neural networks,
• 45:56and you run networks is again an
• 45:59algorithm that allows us to approximate
• 46:02almost any mathematical function
• 46:03that correlate input into output,
• 46:06and it does that by finding
• 46:09interaction between features.
• 46:11Convolutional neural network allow
• 46:13us to search for petitive patterns
• 46:16within an image and then correlate
• 46:18them to the output variable so it
• 46:20tries to find high level features such
• 46:22as edges and the more deeper we go,
• 46:25the more complex the feature become
• 46:28and allow us to separate between the
• 46:31groups that were trying to suffer.
• 46:34A little bit of our data,
• 46:36so we had 120 patients 18 years old or older,
• 46:4088 minutes and 32 females.
• 46:42All of them were diagnosed with
• 46:45HTC between there is 2005 to 2018.
• 46:48So the patient went into MRI imaging,
• 46:52then were diagnosed with HTC and
• 46:54then got treatment 29 oblations,
• 46:5732 receptions and 5:59 presentations.
• 47:01An time went by, some of them recur,
• 47:04and some of them stop their follow which
• 47:07we considered to be non recurrences.
• 47:09To this time that I can call time
• 47:11to recurrence and this would be
• 47:13our explained variable,
• 47:14the variable that we're going
• 47:16to try to predict.
• 47:18With respect to our input data,
• 47:20we're going to use conference enhanced multi
• 47:22phase liver magnetic resonance imaging,
• 47:24MRI,
• 47:25and we're going to use three
• 47:27different phases that the arterial,
• 47:29the portal venous and the delay.
• 47:33So the question that we're trying
• 47:35to answer here is can we predict ATC
• 47:38recurrence using pretreatment MRI images?
• 47:40In other words, are they visual
• 47:42features in free treatment MRI that
• 47:44correlate with HTC disease recurrence?
• 47:46So to visualize that we're going
• 47:48to use the input MRI input data
• 47:50as an input data to freedom into
• 47:52a predictive model which will be
• 47:55convolution on your network and to
• 47:57predict whether the patient will
• 47:59reoccur within one year two years
• 48:01after six years after treatment.
• 48:05So here are the results.
• 48:07Here we can see the results
• 48:09for these 6 * 6 timeframe.
• 48:11So one year for occurrence,
• 48:13two years of free France up to 6,
• 48:16zero for occurrence and this figure
• 48:18present the relationship between
• 48:20the true positive rate to the
• 48:22false positive rate on our test
• 48:24cohort and seeing these 45% curve,
• 48:26which represents basically
• 48:27a random chance to predict,
• 48:29we can see that all of our
• 48:32curves is above that.
• 48:33Which means that our model
• 48:35has prediction power.
• 48:39Another analysis that we did we
• 48:41did was to try to use Kaplan Meier
• 48:44curves and our predictions to
• 48:46separate our cohort into risk groups.
• 48:49Soloist group to recurrence in
• 48:51high risk to recurrence and we
• 48:53got significant results for your
• 48:55current free survival for four
• 48:57out of the six times you can here,
• 48:59you can see the six different figures
• 49:03each for one time for each time frame.
• 49:06So to summarize,
• 49:08the current state of the art criteria
• 49:10based on low level handcrafted
• 49:12radiological features such as tumor size,
• 49:15number of tumors and this study
• 49:17showed that there are still unknown
• 49:20visual features in pretreatment.
• 49:22MRI does correlate with recurrence of ATC.
• 49:24Secondly,
• 49:25the current state of the art selection
• 49:28criteria suffers from false positives
• 49:30and we showed by incorporating
• 49:32machine learning based algorithms
• 49:33we could potentially improve the
• 49:35prediction of AGC recurrence.
• 49:37Little bit of limitations
• 49:40and future research.
• 49:41We used a single site cohort
• 49:44from Yale Hospital,
• 49:46which was small 120 patient,
• 49:48which has the limitation of generalize
• 49:51the results for other sites and
• 49:54therefore will need to increase
• 49:56our sample size or incorporating
• 49:59patient from different side.
• 50:01And maybe single data modality,
• 50:03maybe not enough,
• 50:05so in future research we recommend
• 50:07to test interaction between
• 50:09imaging and clinical data,
• 50:11even incorporating more imaging modalities.
• 50:15Thank you very much.
• 50:23So, so that was really great.
• 50:25Hope you all were able to
• 50:28see what exciting work word.
• 50:30You know, doing in this lab?
• 50:32Does anybody have any questions?
• 50:35And individuals can certainly
• 50:37submit their questions on chat,
• 50:39but we did receive one which
• 50:41David you took the Liberty of
• 50:44at least answering online.
• 50:45But let me for the benefit of all.
• 50:49Let me just toss this out so
• 50:52Joseph Cam asked, you know,
• 50:54essentially, how do you decide?
• 50:56On the specific procedure be RFA,
• 50:59thermal abrasion,
• 51:00ablation, cryoablation,
• 51:01and and also relative to contraindications.
• 51:04How do you make those choices?
• 51:08Yeah, so so. I guess I gave a
• 51:11pretty long email for a chat,
• 51:14but are more common for chat, but.
• 51:17Each ablation modality has its pros and cons,
• 51:21and clearly my goal for today was
• 51:24not to go into my own personal
• 51:27research and discuss a lot of this
• 51:30stuff in depth because I really
• 51:32wanted to highlight our trainees.
• 51:35But basically you know,
• 51:37for liver it's pretty much understood that
• 51:40heat ablation is actually what is preferred,
• 51:43and that's because over the years well,
• 51:46first of all it's very effective.
• 51:49But over the years,
• 51:51there's been concerns with Cryo Ablation on
• 51:54what's called something called Cryo shock.
• 51:57There's questions about fracturing the liver.
• 51:59There's no way of really cauterizing
• 52:02any bleed that you can do with microwave
• 52:06or RFA and a whole bunch of others.
• 52:09So it's really the complications
• 52:12in this in this setting that makes
• 52:16it less ideal than four.
• 52:18For the Heat base,
• 52:20now the one positive thing is we
• 52:23shown in some of the in one of the
• 52:27cases is that you do get the ice
• 52:30ball which has low density ice and
• 52:33that actually allows the operator
• 52:36to really sculpt the margins where
• 52:38you want to treat in other organs.
• 52:41Systems like the lung and kidney
• 52:44is really a dealers choice so that
• 52:47there have been papers on both.
• 52:50In terms of using cryoablation or
• 52:52heat ablation for those that we were
• 52:55just involved in a study that was
• 52:57recently published in the Journal
• 53:00of thoracic oncology,
• 53:01which was a prospective clinical
• 53:04trial called solstice that.
• 53:06Did show benefits of cold,
• 53:08you know,
• 53:09for for lung nodules and then I
• 53:12already as I kind of explained,
• 53:15it's kind of a niche application.
• 53:18Mostly used in pancreas,
• 53:20but it's in an area where it's
• 53:23difficult to treat.
• 53:25Areas which have vital structures
• 53:27in your body,
• 53:29so that's the I guess summary
• 53:31of that so thanks.
• 53:35So let me ask a follow up question 'cause
• 53:38you great deal of the work is really
• 53:42advancing understanding the nature of
• 53:44the immune infiltrate post procedure.
• 53:47And Moreover, how to address how to
• 53:50make the environment more hospitable?
• 53:52Two immune cells, Sony aspid,
• 53:54be both with regard to the imaging
• 53:58techniques monitoring techniques and
• 54:00perhaps the introduction of bicarbonate.
• 54:03How much you leverage all of this to think
• 54:07about combining these approaches with
• 54:09ongoing immunotherapy immuno therapies
• 54:12that are under development in HTC?
• 54:16Right, I mean that's a really
• 54:18important question that we're
• 54:19actually trying to answer ourselves.
• 54:21As David mentioned,
• 54:22I mean, we're right now.
• 54:23Actually submitting one or an
• 54:25application after the other
• 54:26because we want to investigate
• 54:28exactly those points we generated.
• 54:30This preliminary data.
• 54:30Now we want to take it to the
• 54:33clinical perspective and do also
• 54:35trials and immediate future.
• 54:36So I think there is a couple of
• 54:39limitations that we first need to address,
• 54:41and for the pH Image Ng,
• 54:43this is certainly the contrast agent,
• 54:45which is not done yet.
• 54:47Applied and in the human scenario,
• 54:49so we need to make sure that we
• 54:51can do that safely, you know.
• 54:53Secondly,
• 54:54I think you know the imaging of the
• 54:56immune system is a very hot topic and
• 54:58is being hotly discussed and worked
• 55:00up on both by the nuclear medicine
• 55:03community as well as you know us,
• 55:05and I think that both approaches
• 55:06have their merits in their drawbacks,
• 55:08and I think that this will
• 55:10definitely bring a breakthrough
• 55:12in terms of imaging biomarkers.
• 55:13As we know in HC see only 20% of
• 55:16the patients at Prof. Roundabout.
• 55:18Respond to immunotherapy.
• 55:19But we treat a lot of them with it,
• 55:21and so we want to make sure we choose
• 55:24their right patients for the right therapy.
• 55:26And that's the whole goal.
• 55:28As for you,
• 55:29As for the As for the real
• 55:30application of bicarb,
• 55:32that is an approach that could be
• 55:34translated into clinical trial very easily.
• 55:36I mean,
• 55:36bicarbonate is not a harmful substance.
• 55:38It could be used in just used
• 55:40in a clinical trial with taste,
• 55:42and I think we need to really
• 55:44think about using using that as
• 55:46an addition and an additional.
• 55:48Arm,
• 55:48possibly in one of the clinical
• 55:50trials that David is bringing on
• 55:52board to use taste in combination
• 55:54with the immuno therapies.
• 55:55And I think that that would
• 55:57be a very simple initial fix.
• 55:59To do that.
• 56:00I mean there is very little Harmon.
• 56:02Probably you know some major
• 56:03benefit that we could get from that.
• 56:06So we need clinical trials in this
• 56:07respect and we need more translational
• 56:09contrast agent and more research.
• 56:11And I think we do have an
• 56:14amazing infrastructure with RMR
• 56:15Research Center here and the Pet
• 56:17Center on the other hand 2.
• 56:19To actually do that kind of research.
• 56:21And so let me as a follow up.
• 56:23So let me ask, given what you're
• 56:25describing in the nature of the immune.
• 56:27Response, immune environment,
• 56:29fear after procedures such as these.
• 56:34Desert potentially suggests that any
• 56:36efforts right now to combine a checkpoint
• 56:40inhibitor with tastes or other ablation
• 56:43procedure may not be yet optimized.
• 56:46Turn to really achieve what we
• 56:48hope they would achieve because of
• 56:50the nature of the environment absolutely,
• 56:52and this is exactly what I showed.
• 56:55If you remember one of the slides
• 56:57had different embolic agents,
• 56:59and so we're using various symbolic agents,
• 57:01various agents.
• 57:02If we inject into the tumor,
• 57:04that may be pro or anti inflammatory.
• 57:06We then also combine taste and
• 57:08checkpoint inhibition in tumors that
• 57:10may be hot or immunologically cold.
• 57:12So essentially you know a lot of
• 57:14these trials will probably have
• 57:16very non significant results
• 57:18and probably will never achieve.
• 57:20Yeah, it is certain level of clinical
• 57:22translation that we hope for.
• 57:24Just because we don't know what we're doing.
• 57:27Essentially so when we combine Interventional
• 57:28in college and Immuno Oncology,
• 57:30we just go about it and say one size
• 57:33fits all and that is something that I
• 57:35feel like is is the major key point
• 57:38that we need to address and one of
• 57:40these issues is that in Interventional
• 57:42Ecology Community would just did not
• 57:44have the academic culture and you
• 57:46know research environment so far and
• 57:48this is across multiple institutions
• 57:50across our entire community.
• 57:51To actually tackle those topics,
• 57:53and I think now we as we more
• 57:55understand that our therapies are
• 57:57necessary in our increasingly
• 57:58combined with systemic therapy,
• 58:00we need to investigate that,
• 58:01and I think that we're going to
• 58:04have major breakthroughs in the
• 58:05next three to five years. Thank you.
• 58:08Well, I know where at the top of the hour,
• 58:11and I want to thank Jessica and tile
• 58:13and Julius and David for really
• 58:15a remarkably stimulating body
• 58:17of work and Bradley bring to our
• 58:19attention what can be accomplished in
• 58:21Interventional Onkologie So you know,
• 58:23thank you for this continued education
• 58:25for the great work you're doing
• 58:27and everyone on line. Thank you.