Head & Neck Cancers Program 2022: Advancing the Role of Surgery in Head & Neck Cancers

October 13, 2022

Information

October 13, 2022

Presentations by: Saral Mehra, MD, Avanti Verma, MD, and Sam Payabvash, MD

ID8161

To CiteDCA Citation Guide

00:00Welcome everybody to the second
00:02installment of the head and neck
00:05cancer programs 2022 CME Series Co,
00:07hosted by myself and Doctor Burtness.
00:10Unfortunately,
00:11Doctor Burtness can't be here today,
00:13which is fairly appropriate because we're
00:15really talking a lot about advancing the
00:18role of surgery and head neck cancers.
00:20I'll be filling in for Doctor Burton,
00:23speaking about a exciting new trial that,
00:25well, not for us in the academic world,
00:28not so new,
00:29but it's recently been published.
00:31Um, the ECOG Akron 3311 trial.
00:35And so I'll be filling in
00:36for Doctor Burtness on that.
00:38So the this is part two of three
00:41for our head Neck cancer program
00:43at Yale New Haven Hospital,
00:46Smilow Cancer Hospital and Yale University.
00:49We have 3 speakers today.
00:51The first will be me filling in for
00:54Doctor Burtness on the 3311 trial.
00:56I'm talking about D intensification
00:59of radiation therapy for HPV.
01:01Positive oropharynx cancer.
01:03And then we have Doctor Verma who will
01:06be speaking about the appropriate
01:08use of tours and open surgery in
01:11management of head neck cancer.
01:14And of course,
01:14we have Doctor Sam and Payab Avash
01:17who will be speaking about radiomics
01:19of head and neck cancer and I'll
01:21introduce them before they speak.
01:23So without further ado,
01:25I think our numbers are kind of leveling off,
01:28so.
01:29I'm going to get started talking
01:32about this very interesting and
01:35exciting new trial recently published.
01:41So I'm not going to introduce myself,
01:44but there I am, Sir Omara,
01:45I'm associate professor of surgery at
01:48Yale and the chief of Head Neck surgery.
01:51Here we have a wonderful head neck
01:53cancer team and I'm fortunate enough
01:56to lead the surgical aspect of that.
01:59So this trial is entitled Transoral
02:02oral robotic surgical resection followed
02:05by randomization to lower standard
02:08dose IRT for resectable P-16 positive.
02:11Locally advanced oral pharynx cancer.
02:14This is in our circles really known as 3311,
02:18basically ECOG 3311.
02:22And the authors you can see are here,
02:24and I am using some of their slides,
02:26particularly Doctor Ferris from
02:28Pittsburgh, the lead author,
02:29and of course Dr Burtness,
02:31the senior author on the paper as
02:34well in the last last author here.
02:39So this this is a interesting study
02:42because on the population included
02:45P 16 positive newly diagnosed
02:48oropharynx cancer patients who are
02:51amenable to transoral resection.
02:53The treatment was essentially
02:55transoral surgery and then risk
02:58adjusted post operative therapy.
03:01So the risk status was determined
03:04by postoperative pathologic
03:05parameters like extranodal extension
03:07margin status and the number of.
03:09Positive metastatic nodes.
03:10I'm going to show a schema in the
03:13next slide that will really describe
03:16the different groups and how risk
03:18stratification was performed.
03:20In addition,
03:21intermediate risk patients for
03:23subgroup analysis were stratified by
03:25smoking history less than 10 versus
03:27greater than 10 pack years of history.
03:30And in this study,
03:31important functional assessments
03:33were also done,
03:34including modified barium swallows and
03:36patient reported outcomes including the fact.
03:39Your neck and the M daddy dysphagia index.
03:44So this is really the key slide to
03:46understand how this study was conducted.
03:48So that we had HPV P 16 positive squamous
03:51cell carcinoma or the oral pharynx
03:53stage in the 7th edition three or four,
03:56but they were all T1 or T2 and
04:00N1 or 2B in 122B cancers and they
04:04were baseline functional and
04:06quality of life assessments done.
04:08All patients that underwent transoral
04:11resection, this could be laser.
04:14More robotic or bovian headlight,
04:17but they all had transoral
04:20resection and a neck dissection.
04:22Following that patients were
04:24stratified into the low risk arm,
04:26which were negative margins and
04:29no intermediate risk features.
04:31And these patients went on to
04:33observation alone, no radiation,
04:35no chemotherapy.
04:35Then there was the high risk arm.
04:38These patients had positive margins
04:40greater than one millimeter of
04:42extranodal extension or five or more.
04:45Metastatic lymph nodes and they
04:47went on to chemotherapy and
04:50radiation therapy with 66 great.
04:52The randomization actually happened
04:55in these intermediate risk patients
04:58and these were close margins.
05:00The less than or equal to 1 millimeter
05:02VNE and the two to four metastatic
05:04lymph nodes and PN I perineural
05:07invasion and lymphovascular invasion.
05:09These patients were actually
05:11randomized into either 50 Gray
05:13over 25 fractions or 60 Gray.
05:15Over 30 fractions and the outcomes
05:18were two year progression,
05:20free survival,
05:21local regional recurrence and
05:24functional outcomes and quality of life.
05:29There were really two important objectives.
05:32One of them was the feasibility of
05:35doing a multi institutional study
05:38with transoral surgery followed by
05:40risk adjusted adjuvant therapy.
05:43As I can see many of the participants
05:47here know it's pretty challenging to get
05:51surgeons into randomized control trials.
05:54That's sort of the domain.
05:56Generally if of our radiation and more.
06:00Commonly our chemotherapy
06:02or oncology colleagues,
06:03but so one of them was just the
06:05feasibility to do this and we looked
06:07at overall accrual surgical quality
06:08and the risk distribution of patients
06:10that we brought into this study.
06:12The second outcome which I'll be talking
06:15a lot about today is 2 year progression
06:17free survival at 50 Gray versus 60
06:20Gray for those intermediate risk patients.
06:23So can we effectively de intensify
06:25therapy to 50 Gray in these
06:27intermediate risk patients versus 60?
06:30Without impacting 2 year progression
06:33free survival then secondary
06:35objectives were toxicity,
06:37overall survival,
06:38swallowing function and the
06:40patient reported outcomes.
06:42The original study design had
06:45180 called for 180 patients who
06:48were randomized with intermediate
06:50risk and that's assuming that 35%
06:53of patients would be valuable in
06:57that intermediate risk category.
06:59As the study proceeded,
07:00there was a higher proportion of patients
07:02saying that higher risk category,
07:04the RMD where they were getting chemo
07:06radiation and so the total accrual goal
07:08was actually increased to 515 patients.
07:11And there was a plan for interim
07:15analysis at one year for R&amp;B and
07:17CAB&amp;C arms A/B and C and of course
07:20assessing the surgical quality
07:21and risk distribution for the 1st
07:2459 patients completing surgery.
07:29So from this study accrued from 2013 to 2017,
07:34there were 87 credentialed
07:37surgeons and 68 of them accrued
07:39into the study and these patients,
07:41these surgeons perform transoral
07:43resections in 519 P, 16 positive
07:47oropharynx cancers stage T1 to two.
07:51Without matted neck nodes and then
07:53post operative management was
07:55determined based on the risk factor.
07:57So arm A which was observation alone
08:01enrolled 38 patients and then arm D
08:05which was chemotherapy plus radiation,
08:08the high risk patients enrolled
08:11113 patients and then ARM B,
08:14these were the patients that
08:15were randomized to 50 or 60 Gray
08:18enrolled 100 or 109 patients.
08:21And then as I stated before,
08:23ARM D assignment was based on Extranodal
08:26extension more than one millimeter,
08:29greater than 4 nodes and or positive
08:32margin overall in this study the
08:36positive margin rate was 3.3%.
08:40There were some patients that
08:42were deemed ineligible,
08:43which I will discuss briefly as well.
08:46And 27 of those patients had labs
08:49or scans just not done to protocol.
08:52But the treatment arm distribution for
08:54these patients did mirror those for
08:56this 360 eligible and treated patients.
09:01So you can see the reasons for
09:03exclusion from the 519 patients down
09:05to the final group of patients.
09:11And there were a number of reasons.
09:13Some did not receive a transoral resection.
09:16Some patients were just deemed ineligible.
09:18Patients were not assigned or
09:20randomized or never started treatment.
09:22For example, patients who
09:24had end to CN three disease.
09:26And in the end these were the
09:28numbers that I just described.
09:31The reasons patients were ineligible
09:34more specifically were that, for example,
09:37pre study scans or labs were not done within
09:40the four weeks prior to registrations.
09:42If patients had clinical T3 disease
09:44at baseline, they were excluded
09:46or ineligible for this study.
09:48A few patients were unknown while they
09:51were ineligible and to see disease.
09:54The primary was not measurable
09:57radiographically or clinically,
09:58so it could not be appropriately
10:01clinically staged.
10:02There were no nodes at baseline
10:05clinically and then and you can see
10:08the other reasons for ineligibility
10:10in this cohort of patients Step 2.
10:14So step one is a transoral resection
10:16ineligibility and then step two was a
10:19post operative treatment eligibility.
10:21And you can see the reasons for this
10:23for example surgery was performed
10:25more than four weeks from the
10:28registration to step one or.
10:30Starting radiation was greater
10:31than seven weeks post surgery.
10:36The results were actually quite intriguing.
10:38Here you can see the three-year progression
10:42free survival data in that in ARM a,
10:45the low risk group or
10:47there's observational loan,
10:48no radiation transoral surgery alone.
10:50three-year progression free
10:52survival was 96.9%.
10:54The high risk group who received
10:58chemotherapy and radiation
11:00after transoral surgery,
11:02the three-year progression
11:03free survival was 91%.
11:05In these two groups that were
11:08randomized to either 50 Gray or 60 Gray,
11:1050 Gray or 60 Gray,
11:12you can see the three-year progression
11:15free survival was 94.9% and 93.5%.
11:19There were some deaths without
11:21recurrence and you can see here in
11:23the chemoradiation group there were
11:25three deaths in the observation group
11:27there were none and one in each of the.
11:31Randomized groups and you can
11:33see the recurrence numbers,
11:35the absolute numbers as well.
11:39The transoral surgery and low dose
11:42radiation radiation based on this study,
11:45based on these preliminary results is
11:47were there was worthy of further study
11:50based on criteria created a priori.
11:54We also looked at in this study
11:56at the M Daddy scores and the
11:58fact head and neck score,
12:00so patient reported outcomes and what
12:02you can see here is the following.
12:04This is arm a, the M Daddy dysphagia MD
12:09Anderson Dysphasia index composite scores.
12:11You can see the baseline dysphagia
12:14index at 89% post surgery.
12:16There was obviously a drop in the Dysphasia
12:20index and then patients actually recovered.
12:24Quite nicely and you can see the same numbers
12:27here which I'll show graphically shortly.
12:31For arm B, the randomized groups to 50 Gray,
12:3460 Gray and the chemo radiation groups
12:38here and you can see the the decline
12:42in dysphasia index PRO's here in this
12:44in this group of patients and the same
12:48thing was done for the fact head and
12:51neck patient reported outcomes tool.
12:55And the survival curves you can see
12:57for the these are this is for the
12:59eligible and treated patients the
13:00numbers we went over and you can see
13:02how tight they are arm D did have
13:07a slightly lower overall survival,
13:10progression free survival here.
13:13And in the ineligible and treated groups,
13:15this was measured here as well.
13:19And graphically you can see the uh
13:22in the M daddy composite scores,
13:25the observational loan group did best,
13:28RMD chemo radiated did worse
13:31compared to baseline.
13:32But overall these tumors are still quite,
13:34quite, quite good and the intermediate
13:36risk groups and same for the
13:38fact head and neck total scores.
13:40So the conclusion of this study
13:43were the transoral resection for
13:45P-16 positive or famous cancer is
13:48safe and results in good oncologic.
13:50Outcomes and this can offer a promising
13:54de intensification approach to treatment
13:57for oropharynx cancer patients.
14:00In patients who have low risk disease
14:03progression free survival is favorable
14:06without any postoperative therapy and
14:09in those patients who have uninvolved
14:11margins less than five nodes,
14:13minimal or no ENE we can reduce postoperative
14:18radiation therapy without chemotherapy.
14:21Without impacting progression free survival.
14:25So finally transoral surgery with
14:2750 Gray should be in the future
14:31compared to optimal nonsurgical
14:33therapy in some phase three trials
14:37for patients with intermediate risk.
14:42This was coordinated by the ECOG Akron
14:45group here and there were these were
14:48the centers that accrued and Yale was
14:51definitely a major accrued to this study.
14:56I did want to spend a few minutes talking
14:59about two more items related to this study.
15:02One is an abstract that was just recently
15:05presented a few weeks ago at ASCO,
15:08based on data from 3311 not yet published,
15:11but was presented in abstract form with this.
15:17Abstract tried to analyze the
15:20patients from 3311,
15:22looking at patients who smoked 10
15:25greater than 10 Packers or versus
15:28less than 10 Packers.
15:30As most of the people on this call know,
15:32smoking can be a.
15:33A risk factor for worse survival in
15:36oropharynx cancer and HPV associated
15:39cancer puts classically has been
15:42described as an intermediate risk as
15:45opposed to the the high risk patient
15:47and to the favorable risk patients.
15:49This study however showed let me just
15:52get to the the data that there was
15:54no difference in overall survival
15:56or progression free survival for
15:58smokers in this cohort of patients
16:01in 3311 who had transoral resections.
16:03So these even these intermediate.
16:05Risk HPV oral various cancer patients
16:08who are current smokers or who have a
16:10history of greater than 10 pack years
16:12had favorable 3 year progression free
16:15survival and overall survival that
16:16were not worse than those non-smokers
16:19or less than 10 pack your history.
16:21So this data actually shows the first
16:25treatment approach meaning surgery plus.
16:29Radiation therapy without chemo,
16:32in which outcomes were not influenced
16:35by smoking status.
16:37A final study I want to share with
16:38you is tours in the real world.
16:43Where you're treated matters.
16:44This is a study we published a few years ago.
16:47This is actually in 2019 was published,
16:49but I think is quite apropos
16:52to this this discussion.
16:54We looked at the National Cancer
16:57database and looked at positive
16:59margin rates and predictors in
17:01transoral robotic surgery after
17:03federal approval of the robot for.
17:08Oropharynx cancer treatment.
17:11We looked at 3000 patients in the
17:14National Cancer Cancer database
17:16who underwent tours from 2010
17:18to 2014 soon after approval and
17:21we had to exclude some patients,
17:23but ended up with about 2600
17:26patients for analysis.
17:29In the real world during this study
17:32period the positive margin rate was not
17:35the three-point 3% presented in this
17:37study at the at at academic centers,
17:40at credentialed by credentialed
17:42surgeons and it was actually a higher
17:45than a lot of the studies that look at
17:48transoral surgery in at high volume centers.
17:51Nationally the overall positive
17:53margin rate was 17% of patients with
17:56T1 and T2 had a positive margin.
17:59Type of less than 20 percent, 13% and 17%.
18:03And when you get to T3 and T4 cancers,
18:06which I will mention it,
18:08for which the da Vinci robot at
18:10least is not FDA approved,
18:12positive marginal rates
18:14are significantly higher.
18:17In this study,
18:18we looked at factors associated
18:20with positive margin rate and
18:21we found that T classification,
18:23Lymphovascular invasion and volume
18:25of cases by the facility patients
18:29treated at high volume centers were
18:32less likely to yield positive margins.
18:35You can see how we define this was
18:37less than three cases per year,
18:39three to 10 cases,
18:41and then more than 10 cases per year.
18:43And you can see the difference.
18:44A high volume facilities had a rate
18:47still much higher than this study
18:50with credentialed academic surgeons,
18:53but 13% versus 21 percent,
18:5622% for low volume sensors.
18:58So the conclusion of this retrospective
19:02database study was in the year
19:05since FDA approval.
19:06Positive margin rates has been
19:08substantially higher than reported
19:09in high volume tour centers
19:11with academic surgeons.
19:12When you get to higher T stages,
19:16these rates can exceed 28% and then
19:20high volume facilities are half
19:22as likely to yield to positive
19:24margins as compared to low volume
19:26centers on multivariate analysis.
19:29So that's what I wanted to tell
19:31you all about the ECOT 3311 trial,
19:33which basically showed that D
19:36intensification approaches are
19:38possible for HPV associated P-16
19:41positive oropharynx cancer.
19:49Right. So you guys are welcome to
19:53put any questions in the chat.
19:55I'm going to be moderating.
19:57You probably won't see them all,
19:58but I'll I'll moderate them and.
20:01We'll move on to our next.
20:04Next, speakers and we'll do the
20:06questions probably at the end
20:08unless I see something that I think
20:10needs to be addressed right away.
20:13So our next speaker, thanks,
20:16Evan, you must start sharing.
20:18Our next speaker is Doctor Avanti Verma,
20:20who returned to us at Yale after
20:25her years as an undergraduate here.
20:27And even additionally you're
20:29doing research here.
20:30She went off to New York and Atlanta to do.
20:33Our ENT and advanced head and neck
20:36cancer training and we were lucky
20:38enough to recruit her back to New
20:40Haven as into our section of head
20:42and neck surgery here at Yale.
20:43She's assistant professor of surgery
20:46and the lead of head neck surgery at
20:49the VA in Connecticut here as well.
20:53So Doctor Verma will be speaking
20:55to us about the appropriate use
20:56of tours in open surgery.
20:58Thank you so much.
20:59Doctor Verma.
21:00Yes, thank you for the kind introduction.
21:03So I. We'll be speaking about using
21:06transoral robotic surgery and it was one
21:09of the modalities used in ECOG 3311,
21:12but probably the most
21:14prominent one and sort of.
21:16Think of this as an option and
21:18alternative compared to open
21:19surgery and we'll soon learn that
21:22patient selection really matters.
21:24So I will go through that and some of the
21:27technical aspects of the surgery as well.
21:32OK. So you know, just as an overview
21:34of the head and neck anatomy, Umm,
21:37we think about tumors in these in
21:41this whole region as occurring in
21:43different sites and then within sites,
21:45different sub sites. So there's many,
21:47many different sites of the head and neck.
21:49And how we manage a patient really
21:52depends on the the location of the tumor.
21:55And in all these areas there's blood vessels,
21:57lymphatic channels, nerves that we are
22:00trying to preserve the best we can.
22:02Muscles, bone, cartilage, everything.
22:05So you know, anatomic considerations
22:07are very important to us in general,
22:11the principles of head neck cancer
22:13surgery include complete visualization
22:15of the surgical field,
22:17which can be a challenge given that
22:20we're working in in small areas and
22:23with that visualization we want to
22:25achieve on block tumor resection all
22:28in one piece with negative margins,
22:31traditionally margins.
22:325 millimeters or greater and and then in
22:35addition to doing that as best as we can,
22:38we'd like to preserve surrounding
22:41structures that are important
22:43for function of our patients.
22:45So the the gold standard or
22:47traditional or open approaches.
22:49Generally all of these approaches are
22:51transcervical or through the neck,
22:53requiring a neck or facial
22:55incision on the left hand side.
22:57In the oropharynx category,
22:59the the few approaches to the orphans
23:01that were traditionally used for quite
23:04a while are the mandibular automy
23:06approach which requires a lip split
23:09incision most often and you can see in
23:12that cartoon down there that the mandible.
23:15Sort of split open and you
23:16can see this tongue.
23:18The tongue is retracted to one
23:19side and you can see this tongue
23:21based tumor in the visualization.
23:23As I said,
23:23which is important here is
23:25is very good in this case,
23:27but it requires a lot of work and a lot
23:29of potential morbidity to the patient.
23:32Another approach to large tongue based
23:35tumors includes going through the
23:37floor of mouth sling and musculature
23:39there to bring the tongue down into
23:41the neck and so you can visualize
23:43the almost the entire tongue,
23:45essentially the entire tongue.
23:46Through the neck and respect
23:48your tumor that way,
23:49which again is associated with morbidity.
23:52For smaller tumors of the tongue base,
23:55a trans hyoid approach with the
23:58fairing gotami can also be used.
24:01Don't focus too much on the larynx today,
24:03but you know the open approach to the larynx.
24:08The gold standard again is a
24:10total laryngectomy for Laura.
24:11You know especially advanced
24:13stage laryngeal tumors,
24:14partial interjections can be
24:15considered depending on the location
24:17and the stage of the tumor and the
24:20patients comorbidities and those
24:22include vertical partial laryngectomy
24:23is super cricoid laryngectomy
24:25and a supraglottic laryngectomy.
24:27And later in this talk I'm going
24:29to mention that the robot can
24:31be used for the Super Glottic.
24:33Enemy on the right hand side,
24:35the parapharyngeal space usually
24:38requires a transcervical approach,
24:40which can be achieved with mobilization
24:42or excision of the submandibular gland
24:46into the parapharyngeal space there.
24:49Or it can be done in a trans parotid
24:51approach, which requires a facial nerve
24:55dissection and removal and mobilization of
24:58the deep lobe of the of the parotid gland,
25:01which can be quite extensive.
25:04So some minimally invasive approaches that
25:06have come up recently and are robotic
25:09surgery, which I'll focus on today.
25:12And robotic surgery can be used to
25:15access the oropharynx in lieu of those
25:17bigger approaches that I mentioned.
25:19The supraglottic larynx can also be accessed
25:21as well as the parapharyngeal space.
25:24And in that photo below,
25:25you can see that the surgeon is at the
25:28surgeon console controlling the robotic arms,
25:30which are closer to the patient
25:33and there's an assistant.
25:34Of making sure the arms and
25:36the patient are OK.
25:37On the right hand side,
25:38again there's laser surgery
25:40which has existed for longer,
25:41a couple more decades than
25:43robotic surgery which has really
25:45come up in the past two decades.
25:46And there's many kinds of lasers,
25:49but primarily this is used for the
25:51Super glottic larynx and larynx.
25:53It can be used for the oropharynx
25:55as well and and for the trachea.
25:58So for transoral robotic surgery,
26:00when it was first being used for head
26:03and neck and it was really the US side,
26:06that was the model that was being used.
26:08And on the left hand side,
26:10you can see that there's a surgeon
26:12console with sort of those eye
26:15pieces where the surgeon can see
26:17and have a great view of the field.
26:19And then there's the controls there
26:21that you know the fingers go into and
26:24sort of control the robotic arms,
26:26there's petals that provide.
26:28Pottery and a left sided pedal
26:30that controls the camera as well.
26:32So you really have control of of everything.
26:36In the middle is the patient cart
26:38which is basically what's right at
26:40the patient and the arms have trocars
26:43and instruments going through them
26:45that go into the patient's mouth which
26:48you can see on the right hand side.
26:50And then the final component is the
26:53vision cart which is this tower and a
26:56a screen with really high definition.
26:59Images there for the assistant
27:00to be able to see,
27:02and for the scrub tech and anyone
27:04assisting in the surgery to be
27:06able to see what's going on.
27:10More recently, the A new 4th generation
27:13of robot, also from da Vinci,
27:16has been FDA approved for
27:18surgery of the head and neck,
27:20and it's the single port robots.
27:22On the left hand side,
27:23you can see that there's just one
27:25cannula instead of the three,
27:27and there's a camera and robotic arms
27:29that come out through the single cannula,
27:31which measures 2.5 centimeters in diameter.
27:34So it is quite small and the
27:37camera itself has some flexibility.
27:40As you can see in the middle photo,
27:42you can bend, there's certain pose,
27:44you know, we call it the Cobra
27:46pose so that it can sort of bend
27:47to look up or bend to look down.
27:49And the robotic arms have
27:52more mobility in the wrist.
27:54There's much more degrees of mobility there.
27:57So there's usually one arm that has a four
28:01steps to help retract or grab tissue.
28:03The other arm usually has cautery
28:06and even the forceps arm can be
28:09connected to bipolar cautery,
28:10so you can cauterize.
28:12Both arms and then this model of robot
28:15actually has the option of a fourth
28:17arm that you can use however you'd like,
28:21and some of us will put a second
28:23four steps there to keep longer
28:25retraction on tissue if needed.
28:27On the right hand side,
28:29there's a photo of what the port looks
28:31like going into the patient's mouth.
28:33And then there's a retractor.
28:35This is the FKW retractor that's
28:37holding the mouth open and keeping
28:39the tongue out of out of the way.
28:44So the indications for towards Umm,
28:46you know we discussed it a little
28:47bit when we were talking about ECOG
28:493311 on particularly early stage or
28:52financial squamous cell carcinoma.
28:54So T1 and T2 lesions of the tonsil
28:57and tongue base cancers of the soft
29:00palate primarily you know are not
29:02always so amenable to this because
29:05of functional downsides such as
29:07Villa Ferringer and sufficiency.
29:09Isolated lesions of the posterior
29:11pharyngeal wall may be considered,
29:13but if quite a bit.
29:14That is being resected and we usually
29:17do not proceed with this approach.
29:19I mentioned early stage supraglottic
29:21squamous cell carcinoma which
29:23I'll mention again later and then
29:25benign tumors of the oropharynx,
29:27supraglottis and the parapharyngeal
29:30space could also be considered
29:32for transoral robotic resection.
29:35Transoral robotic surgery can
29:36also be used for sleep apnea,
29:37but I I won't focus on that today.
29:40Lingual tonsillectomy or tongue based
29:42reduction can be done for patients who.
29:44To have this contributing
29:46to their sleep apnea,
29:48the first photo is sort of our
29:50view when we have good retraction,
29:53there's a tonsil tumor on the right
29:55hand side and you know the head is
29:57at the bottom of the screen and the
29:58chin is at the top of the screen.
30:00And so we're looking from above and you
30:03know we can see the tip of the epiglottis,
30:06the tongue base on both sides.
30:07We have full view of the tonsil
30:10cancer and even margins of tissue
30:12around it and you can see that.
30:14Four steps in the left hand corner
30:17corner just ready to start.
30:18And then on the right hand side there's
30:21a specimen post resection that I was
30:24trying to Orient for the pathologist.
30:27So using some of the anatomic
30:29landmarks can help because this
30:31is very much A3 dimensional tumor
30:33and A3 dimensional resection and
30:35all of this technology helps us,
30:38you know,
30:38see where we need to see to to achieve this.
30:42So the clinical evaluation for tours
30:44when we see a patient who's referred to
30:47to see if this is even something that
30:50we can offer depend on many things.
30:53The tumor factors include the
30:55size and size is close is closely
30:57in hand with the stage.
30:59So T1 or T2 tumors,
31:01anything bigger than that we
31:03probably would not consider this.
31:05The location is also important
31:08midline tumors sometimes.
31:10Are at risk of injury to both lingual
31:13arteries which I'll mention again later.
31:16So you know we we prefer to do this
31:19approach for lateralized tumors.
31:22The depth also factors into stage and
31:25we can tell that sometimes by palpation
31:28on clinical exam and some you know
31:31it definitely with imaging as well.
31:34If there's trismus this patient can
31:36only open 2 centimeters and and you know
31:39someone's really pushing with their thumb.
31:41You know then that that you
31:43know won't be good for us,
31:45especially with the 2.5 centimeter,
31:47uh,
31:47cannula that it has to get in the mouth.
31:49So that's something that's important
31:51to consider.
31:52The tongue size is also important.
31:55There's a lot of tease here.
31:56So people sort of remember this as the rule
31:59of the teas status of the teeth and the jaw,
32:02whether they're mandibular Tori and then
32:05neck mobility or tilting of the neck,
32:08anyone who's had spinal
32:09instability or spinal surgery.
32:11You know we have to evaluate for
32:13that to make sure we can get
32:15the exposure we need for this.
32:17I added to this prior treatment as well in
32:20patients who have had previous surgery,
32:23previous radiation in particular and
32:25this is sort of a salvage surgery.
32:29We have other considerations including
32:31you know whether that when we do
32:35these resections we don't necessarily
32:37put tissue you know to reconstruct,
32:40but in these patients we
32:41might have to do that.
32:42To protect the carotid artery
32:44or any other vital structures,
32:46because healing might not be as
32:48optimal as As for patients who
32:52haven't been treated before.
32:54So the you know the clinical
32:56exam it you know we rely on a
32:59transoral inspection and palpation,
33:01but we also rely on a flexible
33:04laryngoscopic exam.
33:05And between those two things we
33:07kind of have a sense of the extent,
33:09location,
33:10depth of the tumor whether we
33:12could access this.
33:14But a radiology is also very
33:16helpful for us when we evaluate
33:19patients and and tumors the location
33:22of the carotid artery is.
33:24Is very important,
33:25especially if we're looking at a
33:27patient who has a retropharyngeal
33:28carotid and you can see that
33:30pretty prominently in the left
33:32side in that image with the arrows.
33:34So if this patient needed a
33:37radical tonsillectomy,
33:38we we probably would not consider
33:39that in this case if we thought that
33:42the carotid artery would be exposed
33:44or even potentially injured by this,
33:46by this approach.
33:49Tongue based tumors,
33:50sometimes we can,
33:51we can still do um resections on a
33:54patient with a retropharyngeal carotid,
33:56but that takes a lot of you know
33:59very intense study of the of the
34:01skin and making sure that it it will
34:04stay away from your resection bed
34:06or that you won't keep it exposed.
34:09Then you know the I mentioned
34:10before that the tumor could be
34:12closely associated with the lingual
34:14arteries on both sides and that
34:16would be a contraindication to
34:18transoral robotic surgery.
34:19In the middle image,
34:21you can see a very endophytic
34:22tumor um that is invading and
34:25extrinsic tongue musculature.
34:27So this would already be a higher
34:29stage and a contraindication to tours,
34:31but this likely would involve
34:33the lingual artery on that side.
34:35On the right hand side is probably
34:38a similarly sized tumor but much
34:40more exophytic.
34:41So that's sort of a counterpoint to
34:43that middle image that sometimes we
34:45see these endophytic tumors that
34:47wouldn't be ideal for transoral surgery,
34:49but then sometimes we see these.
34:51Exophytic ones where we know we can
34:53stay away from the lingual artery
34:56and and resect it with minimal
34:59functional morbidity to the patient.
35:01So the the benefits of tours
35:03is avoidance of tracheostomy
35:05which would be necessary in some of these
35:08open approaches and faster rehabilitation,
35:10recovery of speech and swallow function.
35:13The surgery is shorter as
35:15well as the hospital stay.
35:17And we would really consider this as
35:20Doctor Mayer mentioned in his talk if
35:22we were able to reduce or eliminate even
35:25the need for post operative treatment.
35:27So post operative or adjuvant
35:29radiation or chemo.
35:31Therapy and I just put the ECOG
35:333311 schema up there.
35:35Just as a reminder as to you know
35:37one of the reasons we do transoral
35:39robotic surgery to minimize
35:41adjuvant treatment if we can.
35:43If we think that we could not say
35:45there's a small primary tumor,
35:46but there's matted nodes and obvious
35:49extranodal extension that you know that
35:52would put them in in the high risk arm,
35:55then maybe you know we should
35:58consider upfront chemo radiation.
36:01The risk of tours include taste disturbance
36:04and tongue numbness on those 2GO hand
36:07in hand primarily because of retraction.
36:10Using these FK retractors or the Med
36:12robotic retractors or even the Crow
36:14Davis and having that that blade blade
36:17against the tongue for quite a while
36:19can cause these things and and some
36:22patients last longer than others.
36:24Even though one of the things
36:26that is improved with Torres is
36:28swallowing return to swallow function,
36:30there still can be problems with
36:31swallowing in the immediate.
36:32Post operative period Velopharyngeal
36:35insufficiency is something that I mentioned,
36:38particularly if there is a significant
36:40portion of the soft palate that's
36:43resected if in in oropharyngeal
36:45squamous cell carcinoma in particular,
36:47a neck dissection typically is performed
36:51with the transoral robotic surgery.
36:53And so when that happens,
36:55there is a risk of fistula between
36:58the oropharynx resection and the neck
37:00and so we have to monitor for that.
37:03Carefully, um,
37:04during the surgeries in some cases,
37:06in some surgeons would tend you
37:08know would tend to stage the neck
37:10dissection for do that first and
37:12then do the robotic surgery later
37:14or if we do it at the same time,
37:17there are some local reconstruction
37:19options that we can consider to
37:21close a fistula and you know those
37:24that's important to keep an eye out
37:26for the other thing about the neck
37:29portion of the surgery particularly in
37:32oropharyngeal squamous cell carcinoma.
37:33Is ligation of the external
37:35carotid artery or its branches.
37:37Because bleeding is a very
37:40feared risk of tours,
37:41it can be life threatening and
37:44it doesn't happen immediately.
37:45It tends to happen over
37:47a week postoperatively.
37:49So it is standard of care to ligate these
37:53branches to minimize the risk of bleeding.
37:58Contraindications to transoral robotic
38:00surgery include inability to visualize
38:02the lesion or any relevant anatomy.
38:05Trans orally if there is carotid
38:07artery involvement of the tumor which
38:10would upstage the tumor as well,
38:12prevertebral fashion involvement,
38:14any mandibular invasion,
38:16and if there's greater than 50% tongue
38:19based involvement or greater than 50%
38:21posterior pharyngeal wall involvement.
38:23I also mentioned the medicalized
38:26or retropharyngeal carotid.
38:28Um, which is generally a contraindication.
38:31Um, but sometimes it particularly
38:33in tongue based resections.
38:35If it's really posterior,
38:37it can be still considered.
38:41So this is sort of some
38:44relevant internal anatomy.
38:45The the two pictures on the
38:48left depict the initial approach
38:50to a radical tonsillectomy.
38:53So the forceps is holding the
38:56pharyngeal constrictor muscle and this
38:58is sort of the first incision that
39:00we make in a radical tonsillectomy.
39:02We expose that medial teratoid muscle
39:05and and use that pharyngeal constrictor
39:07as as our deep margin essentially.
39:10And as that's being retracted medially
39:12and that sort of whitish fluffy thing
39:15stuff you see is the parapharyngeal fat.
39:18And so that's where the,
39:20the blood vessels are the things
39:22you want to avoid.
39:23And so that sort of bluntly gets.
39:27Dissected laterally so we can
39:29continue working on the muscle
39:31that can get transected and often
39:33does is the styloglossus muscle
39:35where that blue arrow is.
39:37So there's a lot of internal
39:39anatomy that we're thinking about
39:41as we do these resections and on
39:44the right hand side that's that's
39:46a depiction of the location of the
39:48lingual artery because during a
39:50tongue based resection or focused
39:52on where that is and you can see
39:55that that dorsal lingual artery.
39:57In in the circle there that that's
39:59often that often can show up and we
40:01are watching out for it especially
40:03it almost comes up as like a
40:05knuckle of a vessel as you're in
40:07the tongue based musculature.
40:09So we're always watching out
40:10for that when we do these,
40:12do these cases and thanks to the
40:15technology of the robot we really have
40:18great visualization as we're working.
40:20So moving on to Super Glottic squamous
40:23cell carcinoma exposure is still important.
40:26Sometimes it can be more
40:28challenging in in in that photo,
40:29some of those longer blades are used to
40:32get deeper and exposed to super glottis.
40:35Again,
40:35transoral robotic surgery is used
40:38for early stage tumors at T1 and T2.
40:41Some surgeons do report using
40:43this for T3 tumors as long as
40:45both vocal cords are mobile,
40:47which basically precludes paragliding.
40:50Face involvement.
40:52The airway considerations you know
40:54are interesting and super chaotic.
40:57Squamous cell carcinoma.
40:58I think there are some cases in
41:00which doing a tracheostomy up front
41:03to protect the airway and and in
41:05the event of any bleeding could
41:07be considered more so than for or
41:10pharyngeal squamous cell carcinoma.
41:12And the major source of bleeding
41:14if it were to occur would be from
41:16the superior laryngeal artery in
41:18this case and in the bottom photo.
41:20You can see that there's on the
41:22lingual surface of the epiglottis,
41:23there is a tumor there and so
41:26there the robot is being used to
41:29sort of visualize it and also then
41:32for resection contraindications
41:34to this include limited exposure.
41:37Poor pulmonary reserve is actually
41:39a contraindication to a supraglottic
41:41laryngectomy as well as you know
41:43it can be a difficult recovery,
41:46there can be aspiration postoperatively.
41:50Involvement of the anterior commissure
41:52thyroid cartilage is also contraindication,
41:54as is periodic space invasion.
41:56That could cause vocal cord fixation
42:00or a hypomobility.
42:03Some tumors that with minimal
42:04involvement of the pyriform
42:06sinus can be resected this way.
42:07But if there is involvement of
42:09the apex of the pyriform sinus
42:11or any post cricoid mucosa,
42:13that's also a contraindication.
42:17And just briefly for the final slides,
42:19wanted to review trends or robotic
42:21surgery for the parapharyngeal space.
42:23Generally it's used for a well
42:26circumscribed tumors most commonly
42:28for a deep lobe parotid neoplasm such
42:31as a pleomorphic adenoma and these
42:34generally are in the pre styloid space.
42:37And the the bottom picture if you
42:39can see it sort of depicts the pre
42:42styloid versus the post styloid,
42:44but generally the pre styloid
42:46space is occupied by.
42:47Parapharyngeal fat in the post dilate
42:49space is where the great vessels are
42:52and and and the associated nerves.
42:54So This is why we would probably not
42:57use this for post thyroid tumor.
43:00There have been reports of resecting
43:03benign tumors as large as 8
43:06centimeters trans orally there are.
43:08I think again selection is very important.
43:11The relationship to the internal
43:13carotid artery is key here.
43:14So if the artery is displaced
43:16laterally and you can see it.
43:18A plane between the tumor and the carotid,
43:21I think you know that that's a
43:23sign that this could be safe to do
43:25with a transoral robotic approach.
43:27In some cases with these deep lobe
43:30salivary gland tumors or prodded
43:31gland tumors,
43:32extension through the stylo mandibular
43:35tunnel may require a combined or
43:38an open approach because of that.
43:41Because it's hard to get laterally
43:43beyond and I'll show you photos
43:45of that and that this creates a
43:47dumbbell appearance on imaging.
43:48So the the graphic on top just
43:50shows where the stylo mandibular
43:52ligament is and that barrier sort
43:55of causes the tumor to grow around
43:57it and create a dumbbell.
43:59And so part of that could be accessed
44:02very easily trans orally as you
44:04can see the bottom right image.
44:06But the part that's abutting the deep
44:08lobe of the product can be difficult.
44:10So sometimes this.
44:11This would require just an open
44:13approach or or a combined approach.
44:16The advantages of of going trans orally
44:19to approach these is that there's less
44:21risk of first bite syndrome and fry syndrome,
44:24which are well described after
44:27transcervical or Transpara added
44:29approaches to the parapharyngeal space.
44:32There's less risk of Cialis steel because
44:34you're not dissecting the parotid gland,
44:36and then there's less risk of facial
44:38nerve injury because you're you're
44:39not dissecting that either.
44:40And of course there's no external incision.
44:45That,
44:45you know,
44:46the main disadvantage is that there's a
44:48very narrow corridor of exposure and if
44:51there's any bleeding that occurs there,
44:54there can be a lot of
44:56difficulty controlling that.
44:56So sometimes when when surgeons
44:59are doing this,
45:00they'll have a backup open approach
45:02so you can convert to an open approach
45:05if needed to control any bleeding,
45:07even though there's a decreased
45:09risk of facial nerve injury and
45:12other adverse effects,
45:13the glossopharyngeal nerve.
45:14Is actually closely associated
45:16with the styloglossus muscle,
45:18so that's an increased risk of injury here.
45:22So the you know in conclusion traditional
45:24open surgery is the gold standard
45:26and and what's been done for many,
45:28many, many years to approach the tumors
45:30of the head and neck with the following
45:33purposes to visualization on block
45:35resection with negative margins and
45:37preservation of surrounding structures.
45:39Robotic surgery over the past
45:41two decades or slightly more has
45:43demonstrated that you know we can
45:45achieve similar outcomes with improved
45:47functional outcomes but we have
45:49to select our patients carefully.
45:51Based on clinical aspects and
45:54radio graphic aspects too.
45:58So here are my references and I
46:01had a question slide while we were
46:03waiting till the end so. Thank
46:05you. We'll wait till the end. That's great.
46:08Thank you so much Doctor Verma for
46:11that great run through a transoral
46:14robotic surgery and Susan head neck.
46:18Great. So our final speaker for
46:20this evening is not a surgeon,
46:23but we work very closely with him as
46:26surgeons and he's done some really
46:28exciting work on road radiomics.
46:30Dr Pavish is did his MD at Tehran University,
46:34was a research fellow at Mass General,
46:37went through his radiology training and
46:40did a neuro Neuroradiology fellowship,
46:44highly coveted at UCSF.
46:47Probably more than five years
46:48ago at this point.
46:49And now he's assistant professor
46:52at of Radiology in Neuroradiology
46:55here with us at Yale University.
46:57And he's going to be speaking about
46:59Radiomics, so I'll hand it over to him.
47:03Um. Thank you very much.
47:06Thanks Doctor Mehta for
47:07introduction and invitation.
47:09Um, so I will be speaking about
47:14Radiomics in head and neck cancer.
47:19Do not have any conflict of interest
47:23and the talk will be focused on
47:26the application of radiomics.
47:30A short and brief description of
47:33what we are talking about when
47:35we are referring to radio mix,
47:38how this can help with diagnosis
47:42and molecular subtyping of tumors,
47:45prognostication,
47:46prediction of survival and perhaps
47:49treatment planning in patients
47:51with head and neck cancer.
47:53And this was a review article that
47:56we published with Doctor Burtness
47:58a couple of years ago. And so.
48:03When we talk about radiomics,
48:07this basically represent a hard coded
48:12series of hard coded algorithm that
48:17extract numeric features from medical images.
48:22So it was basically started
48:26along with the omics spectrum,
48:29as you might have heard about genomics,
48:33proteomics.
48:34The idea is that we extract a large
48:38amount of numeric and quantitative
48:42information from medical images
48:45and try to harness information
48:48from them for precision diagnosis.
48:52And precision treatment planning.
48:55Now the RADIOMICS features or the
48:58radiomics numbers are in generally
49:00representative of the intensity,
49:03shape and texture of a target lesion.
49:07In this case head and neck cancer intensity,
49:12basically the brightness of the tumor
49:15or lesion of interest on medical images.
49:20We are pretty much always
49:22working with grayscale images,
49:24Umm, the shape of the tumor.
49:26And also the texture,
49:28how much it is heterogeneous and this
49:31large amount of information that we
49:34extract or actually well suited for
49:37machine learning algorithms because
49:40those are preferred and suitable
49:43statistical models to make a prediction.
49:50So some of the references that I make
49:52are related to brain tumors, but you can.
49:57Basically apply the same
49:59concept to head and neck tumors.
50:01So when we talk about intensity features,
50:05you can think about the mean or
50:08range of the intensity or brightness
50:11that you see in a specific lesion,
50:15but it can also get a little
50:18bit more sophisticated.
50:20We can think about the magnitude of the
50:24changes of the voxel values in an image.
50:27Now would be referred to as energy
50:30or entropy like or like randomness
50:32of the values and the image.
50:35So that's why, you know,
50:36we basically get a larger number of
50:40numeric values that are representative
50:44of the intensity feature and.
50:48In medical images,
50:49when we talk about the radiomics
50:52intensity feature, this is again.
50:56Course we did with posterior
50:58fossa tumors and this was like the
51:01information from the histogram,
51:03ADC histogram in these tumors and you
51:06can see how this is different from
51:09one tumor subtypes to another and
51:11we apply this for differentiation of
51:13this posterior fossa brain tumors.
51:16The shape of a tumor may also have an impact.
51:20We usually think about the volume
51:22how big a tumor is but sometimes
51:24the surface and it's.
51:26We may also have an impact.
51:29I haven't found like a good example
51:31in terms of head and neck tumors,
51:34but for example.
51:37This paper which was done on
51:41glioblastoma showed that how much
51:44did the basically minimum volume
51:46of a bounding ellipsoid may have
51:50an impact on the overall survival
51:53of the glioblastoma.
51:55So there is some work that can be done
51:58on looking into how does the shape
52:01of a tumor affect the prognosis or.
52:08Perhaps the way affect treatment planning.
52:12And then the last thing that I
52:14mentioned or the texture of the tumor,
52:15how how much the tumor is heterogeneous.
52:18Now the numbers or the metrics that we
52:21use for this are a little bit complex.
52:24But in general you can think about
52:26it as we are looking into seeing
52:29how much the intensity of 1 region
52:33is different from the region next
52:35to it kind of the same concept that
52:39the more heterogeneous tumor is we
52:41perhaps expected. Could be more advanced.
52:44It's perhaps has like more time to grow.
52:47There are some areas that have necrosis,
52:49some are still, you know,
52:51growing and some are more vascular.
52:55And this was the work that was done
52:57for example for subtyping of the
53:01medulloblastoma along the same line.
53:04When we are looking at the texture
53:06or heterogeneity of the tumor,
53:09we can apply some filters and this
53:14is like a example of how does these
53:17filters change the original image.
53:20For example we can pass,
53:21we can apply low pass and High
53:24Pass filter a low pass.
53:25Winter kind of smooths out the image,
53:28gets the overall view of the what
53:33the original vision is by in a
53:36high pass filter you can look more
53:39into the contrast or the edges
53:41of the image. So.
53:43This is, for example, a prostate cancer.
53:46This is how we apply these filters
53:49in three directions and we get
53:52eight different derivatives.
53:55Again, these are all sorts of
53:57manipulation that we do just to get
54:00more and more information about the.
54:03Are the tumor going above and
54:07beyond the intensity and shape and
54:10specifically trying to figure out
54:12what we can get in terms of the
54:14information from the heterogeneity
54:16or the texture of the tumors, so.
54:20I just referred to like.
54:23Think that you know like three
54:25of the works that we have done
54:27here could showcase of what we
54:29can achieve with radiomics.
54:30One as I mentioned is
54:34molecular subtyping of tumor.
54:37So as you all know HPV status is very
54:40important in terms of prognostication.
54:43It's indeed the first step for
54:46us to decide how we're going
54:48to stage a tumor after 2018.
54:53Adjustment of the AGC.
54:55So we tried to use radiomics features
54:58from Pet city to predict the HPV
55:02status of head and neck tumors.
55:05So we segmented the the.
55:11The primary lesion is segmented the
55:14metastatic lymph nodes on pet CT
55:18we extracted it roughly like 1000.
55:21Features are representing the intensity,
55:24shape and texture of these primary
55:28lesion and tumors we were using.
55:34Roughly A144 from the Cancer
55:38Imaging Archive and 291 from Yale,
55:41and we split it into a training slash,
55:46cross validation and an
55:49independent validation cohort.
55:51We were trying to see which image
55:55modality and which combination of
55:58the input variables are important for
56:01are are would be most accurate in
56:04terms of prediction of HPV status.
56:07So here you can see that it
56:09was very extensive work.
56:10We were trying to see whether pets alone,
56:12city alone or pet city
56:16information using primary tumor.
56:20Lymph nodes or the consensus of the
56:23tumor and nodes or consensus of all lymph
56:27nodes will be give us the best model.
56:30Long story short, is that what?
56:33We found that a combination of the
56:36pet CT using the consensus of the
56:40primary lesion and the lymph nodes
56:43can give us the best prediction.
56:46And these are the AU says that
56:49we could get in our independent
56:52and external validation cohorts.
56:54So you may question that, OK,
56:58so how does this really affect our,
57:03you know, staging really,
57:05because you always will have
57:07a tumor sample to decide.
57:09Now if you look at how the
57:11pathologists actually do it,
57:13there are different stages sometimes,
57:16well,
57:16technically the guideline from
57:19the American Pathological
57:20Association is that they first do.
57:24And immunohistochemistry and then they
57:26do if depending on how certain they
57:29are based on the IHC, they do the PCR.
57:32Here at Yale we pretty much go
57:35for everything we go for PCR.
57:37However,
57:38what we proposed in our paper is that
57:42this is not a substitute for tissue sampling,
57:46but this can mostly work as an adjunct
57:49to the tissue sampling results.
57:51In other words,
57:53if you have a PC order or even
57:59histochemistry that is equivocal,
58:02maybe we can use this to supplement that.
58:07Analysis and that pathology report, so.
58:11Again,
58:12something that is perhaps not going
58:14to at this point we are still far
58:18away from replacing tissue sampling,
58:21but perhaps we have some quantitative
58:25and reliable and reliable methods
58:29to supplement those whenever needed.
58:33Now how about prognostication?
58:35So we try to see whether or not using
58:41the RADIOMICS features can help
58:43with prediction and prognostication
58:45and prediction of the survival
58:48beyond the JC staging.
58:50And the reason why we looked at
58:53a JC staging was because.
58:56It's kind of like you can say like
58:59the benchmark for a prognostication.
59:02So we use the HCA is addition.
59:06We were again using a series of
59:09HPV positive and HPV negative
59:11patients and our modeling was to
59:15predict those both progression for
59:18survival and overall survival
59:21using these radiomics features and.
59:24This is,
59:25I think it can give you the gist of it.
59:28So these are a different time points,
59:312 year, three-year, four year and five year.
59:34And as you can see for both HPV
59:38positive and for HPV negative,
59:41the RADIOMICS features could
59:43differentiate between high risk and low risk.
59:47Uh, patients? Fairly well.
59:50As you can see, uh,
59:52for HIV positive in all four time
59:55points that we
59:57tried, we could achieve significance.
01:00:00P value for differentiation,
01:00:03but really the agency is staging.
01:00:08Was not able to differentiate the
01:00:10low risk and high risk patients.
01:00:13Uh with this I mean even achieving
01:00:16the statistical significance,
01:00:18I should note that we actually excluded
01:00:20any stage four patients from our
01:00:24analysis and the same thing as you.
01:00:28This was actually, I'm sorry,
01:00:30this was an HP negative series, ohh,
01:00:33sorry, this is the overall survival,
01:00:35this is the progression free survival,
01:00:37so kind of. Same story here.
01:00:41Could be I forgot to include this
01:00:43slide about the HPV negative series.
01:00:46But the bottom line is that we and
01:00:51the way we envision this is that in
01:00:54future in addition to just tumor size,
01:00:59lymph node size or the anatomical
01:01:02extension of the tumor,
01:01:04we probably can get a bunch
01:01:07of numbers that can help us.
01:01:10Bitter stage, the patient.
01:01:11This is based on the very baseline pity.
01:01:14This is how currently we proceed
01:01:19to stage our patients.
01:01:21So if we have a better way of staging
01:01:25them at the baseline in terms of
01:01:28the survival and prognostication.
01:01:31Then we will have better way of
01:01:34treatment planning and smarter
01:01:36way of treatment planning, so.
01:01:38I we envision that perhaps in New
01:01:42York near future in addition to.
01:01:46General you know like a staging
01:01:47numbers we may have like more
01:01:49sophisticated numbers that can
01:01:50tell us this is a low risk,
01:01:52this is a high risk patient in terms
01:01:55of the survival and then finally.
01:02:00We did look to see if there are.
01:02:04If Radiomics can help predict locoregional
01:02:09progression after radiotherapy in
01:02:13HPV associated oropharyngeal cancer.
01:02:18And this is a very good follow up
01:02:21to presentation by Doctor Mehra
01:02:23and the E 3311 in the sense that
01:02:29these patients are potential.
01:02:33Candidates for intensity reduction
01:02:36in terms of radiotherapy.
01:02:38So if we know that who are more at
01:02:42risk of post radiotherapy regional
01:02:46progression and who is less likely
01:02:50to have the original progression,
01:02:53then you can perhaps use that
01:02:56for treatment planning.
01:02:59We use kind of similar methodology
01:03:02that we use for survival prediction,
01:03:05this time only focused on HPV positive
01:03:09patients would receive radiotherapy and
01:03:12as you can see here we could basically.
01:03:21Predict the overall survival
01:03:23progression for so and local regional
01:03:26recurrence Indian which is better?
01:03:30Accuracy compared to the agency.
01:03:34If we want to use again agency
01:03:37aging as a prognosticator,
01:03:39one thing that I should mention is
01:03:42that you may see that, you know,
01:03:44like over time we kind of lose the accuracy.
01:03:48It's simply because we had.
01:03:51Smaller number of patients who
01:03:53were followed beyond three years.
01:03:56So when you have less data
01:03:59point your your model,
01:04:00your machine learning model just
01:04:03would not have enough input to
01:04:05generate good prognostic model.
01:04:09So in general this is basically a.
01:04:17A very detailed information detailed
01:04:21of the local regional recurrence,
01:04:24local regional preparation
01:04:26and based on the AGC staging,
01:04:29different stages, oral stage,
01:04:31the age of the patient and
01:04:34how did they reconcile.
01:04:36So in general, Radiomics offers an
01:04:40automated way of image analysis.
01:04:43It will provide a numeric numbers
01:04:47and quantitative metrics for
01:04:49machine learning algorithms.
01:04:52And I tried to present some of the
01:04:55work that we have done here as how
01:04:59we can use this for differentiation,
01:05:03molecular subtyping of the tumors,
01:05:06prediction of the treatment response
01:05:10and survival prognostication.
01:05:15So yeah, let's hope that it
01:05:18was helpful for you. Thank
01:05:21you Doctor Pravesh, that was.
01:05:23Really exciting stuff.
01:05:27There are, there are some questions.
01:05:30Thank you everyone for staying on time.
01:05:32We're pretty much right on time where
01:05:34we want it to be which is great.
01:05:36There are some questions that
01:05:38people had for the first talk there
01:05:41was a question but are we doing
01:05:44deep intensification already?
01:05:47I'll answer that one is the
01:05:49short answer is yes.
01:05:50You know a lot of the as you could
01:05:52see there were I don't know maybe 5060
01:05:55academic centers I'm recruiting to the.
01:05:58He called 3311 trial and you know we
01:06:01were seeing results you know before
01:06:05publication and patients were asking
01:06:08and so off trial you know there
01:06:10was some discussions about this
01:06:12and now once the abstract came out
01:06:14did intensification is happening.
01:06:16I mean some might say tours alone
01:06:18is densification but also deep
01:06:21intensification of the dose of
01:06:23radiation is happening already
01:06:25at at major academic centers.
01:06:27But I think it has to be done in a mindful.
01:06:29Thoughtful way, multidisciplinary way,
01:06:31you know,
01:06:33with with all options presented to patients.
01:06:36Doctor Verma,
01:06:37there was a question about
01:06:39tonsils versus base of tongue.
01:06:42Is 1 easier to approach than the other?
01:06:46Does it impact your decision
01:06:47of what you know?
01:06:49How does that impact your decision
01:06:50to do tours if it's in the base
01:06:52of tongue versus the tonsil?
01:06:56Yeah, that's a good question.
01:06:57I think considerations might
01:07:00be different and particularly
01:07:02the anatomical considerations.
01:07:04If it's a tonsil tumor,
01:07:05you already know it's lateralized,
01:07:08a tongue based tumor.
01:07:09You still have to evaluate.
01:07:10If it's approaching midline for example,
01:07:13then you know we would probably
01:07:16not recommend doing a transoral
01:07:18resection also because we would have
01:07:20to consider management of both sides
01:07:23of the neck in terms of potential.
01:07:25Regional metastasis to lymph nodes.
01:07:28But you know I think if it's,
01:07:31it's not that we would choose
01:07:32one over the other necessarily.
01:07:34I think it's just different considerations,
01:07:36but that's a great question.
01:07:38And then there was a question
01:07:39which you kind of answered in that
01:07:40is when do you do bilateral neck
01:07:42dissections along with tours and how
01:07:43does that factor into your decision
01:07:45of whether or not to do tours?
01:07:48Yep. So yeah bilateral neck dissection
01:07:50would most would probably not be
01:07:53considered in a well lateralized
01:07:54tonsil tumor which inherently is that.
01:07:57But in a tongue based tumor that
01:07:59you're doing it towards resection on
01:08:01and there is approach even you know a
01:08:04couple millimeter or millimeter or so
01:08:07between you know close to midline we we
01:08:09have to consider you know management
01:08:11of both necks and this is actually
01:08:13where again the multidisciplinary
01:08:15approach really matters and before
01:08:17we consider or proceed with this.
01:08:19Kind of surgery,
01:08:20we have a radiation oncology and
01:08:22medical oncology colleagues see the
01:08:25patient and we could consider either
01:08:27not doing the transoral resection and
01:08:30doing chemo radiation or if there's
01:08:32some factor that you know really
01:08:34pushes us towards transoral surgery,
01:08:36we could consider bilateral neck dissections.
01:08:40Great. And then Doctor Paul Bashere was
01:08:43a question about, well, first of all,
01:08:46someone coming is very exciting
01:08:47how you can use radiomic data to
01:08:50help with prognostication and very
01:08:53interesting to hear your reviews about.
01:08:58You know one day potentially putting in it
01:09:01into something like a staging system even,
01:09:03which is really exciting.
01:09:04I mean right now we use pretty crude metrics
01:09:07on imaging like invasion into this muscle.
01:09:10Therefore it is this stage,
01:09:11but you're you're proposing in
01:09:14the future to use on almost,
01:09:16you know numeric data from RADIOMICS
01:09:19to help with pronunciation.
01:09:21Is that is that a true statement
01:09:24and I just want to add that, for example.
01:09:29Right now the Umm, we are working again.
01:09:35It's kind of pioneered by
01:09:37my colleague Dr Maria Mboya,
01:09:40who is working on brain tumors.
01:09:41We have already implemented the.
01:09:46The pipeline that extracted
01:09:49radiomics numbers.
01:09:50So technically speaking
01:09:51she has like an automated.
01:09:54She also had work on an automated
01:09:56segmentation of brain tumors, so.
01:09:59Umm, from the packs,
01:10:02like from the visage packs that, yeah,
01:10:04you can directly get the numbers,
01:10:07the radio mix number for brain tumors.
01:10:09So we can, you know, technically easily
01:10:12apply this to your or pet cities.
01:10:19And get those numbers and we talked about.
01:10:25We literally talked about the
01:10:27models that I developed.
01:10:28It's just that I use.
01:10:30Our coding for um,
01:10:32the machine learning algorithms
01:10:34and the the they were you know.
01:10:40The our practice system has a a Python
01:10:44on basically language compatibility.
01:10:46But yes, I mean literally here we
01:10:51are very close to you know, getting
01:10:55those numbers on on on our tax system.
01:10:59Great. That's wonderful.
01:11:01And then there's another question
01:11:03about how about using radiomics
01:11:05to predict extranodal extension
01:11:07in a lymph node in the neck,
01:11:09which could totally change whether
01:11:12or not we recommend surgery or not
01:11:15based on the current NCCN guideline,
01:11:18treatment recommendations and
01:11:19what are your thoughts on that?
01:11:21Yeah, so actually. Benjamin Connor,
01:11:26who is a radiation oncology resident,
01:11:30he's now at the Dana Farber's.
01:11:32He developed that they use
01:11:34a deep learning model.
01:11:35We actually even tried radomes which
01:11:38was creating like a similar accuracy.
01:11:43And at the time, yeah,
01:11:46like he was about to leave Yale.
01:11:48We talked about bringing in his
01:11:51model to in our park system.
01:11:53It didn't work.
01:11:54And now he's at Dana Farber.
01:11:56They also have like
01:11:58similar pack system as us.
01:12:00I know that he's working on
01:12:02it and his model has like an
01:12:05accuracy close to 0 point like
01:12:0780% and it was even more accurate
01:12:10than radiologist for prediction
01:12:12of the extranodal extension.
01:12:15So I do believe that we are very
01:12:21close to really implementing all of
01:12:23these in our clinical day-to-day.
01:12:27And there's one more
01:12:29question for you, Professor.
01:12:30There are groups of patients
01:12:33for whom radiomics or machine
01:12:35learning models are not,
01:12:37are not as predictive that
01:12:38you just know up front.
01:12:40So we are our models are as
01:12:43good as the data that we have.
01:12:46So that's why I mean when I
01:12:48mentioned that for example we do
01:12:49not have if we don't have data
01:12:51for you know long term prediction
01:12:53or models are not working.
01:12:55So we barely have data on HPV.
01:13:00Negative patients.
01:13:01So our models are less accurate for HPV,
01:13:06negative or for angio cancer.
01:13:10It's. It's only works if we have
01:13:14enough data and yet for example.
01:13:18Since the introduction of PD1 inhibitor,
01:13:22the two the treatment response has changed.
01:13:25So now we have to train a new set
01:13:28of models for prediction of how,
01:13:30how this is gonna you know like
01:13:33how how would that affect the
01:13:35the survival of the patients.
01:13:37So the the all the models that we
01:13:41developed were based on information that
01:13:45came from somewhere from 2011 to 2.
01:13:491016 so there is a lag between the
01:13:52models of your developing and the
01:13:56current cutting edge treatments.
01:13:57We we have to retrain the models
01:14:00based on the the most up-to-date
01:14:03treatments that we have.
01:14:08Very interesting, very interesting.
01:14:09All right. Well,
01:14:10I think that answers all the questions.
01:14:13I wanted to thank our speakers,
01:14:15doctor Avanti Verma,
01:14:17doctor Sam Kavesh for joining us and
01:14:20we want to thank all the participants
01:14:22who logged in today and the ones who
01:14:26will also see this on the website.
01:14:28This will be posted on the
01:14:31Twitter for Yale Cancer Center,
01:14:33will be on the Yale Cancer website and.
01:14:36Please reach out to if
01:14:38you have any questions.
01:14:39Thank you all very much for joining us.
01:14:41Thank you. Thank you.