Precision Medicine vs Persuasion Medicine: Measuring vs Reading HER2

February 22, 2023

Yale Cancer Center Grand Rounds | February 21, 2023

Presented by: Dr. David Rimm

Information

ID: 9544
To Cite: DCA Citation Guide

Download Transcript

00:00I'm, I'm iron crop.
00:02I'm the CTO director here and it's
00:06my pleasure to introduce David Rimm,
00:08who has many of you know is the Anthony Brady
00:11Professor of Pathology and medicine here.
00:14He is a Hopkins alum, did an MD PhD there,
00:18then came here for for pathology residency
00:21and then did a psychopath fellowship
00:24at the Medical College of Virginia.
00:27He's actually now been at
00:29Yale for almost 30 years.
00:31That's impressive.
00:32He's the director of the Pathology
00:35tissue service here and serves as
00:38director of Translational Pathology.
00:40You know, I think David has been
00:42at the forefront of.
00:44I guess I don't need to as in the
00:47forefront of quantitative pathology
00:49for for many years and he's well
00:53known throughout the the, the.
00:55The field for this he developed
00:57many novel assay techniques for
01:00identifying predictive markers to
01:02determine which tumors are sensitive
01:05to which targeted therapies.
01:07And this has become increasingly
01:09important as the number of targeted
01:11therapies has increased and our
01:12use of those drugs has increased.
01:14And today he's going to focus on the
01:16development of companion diagnostics
01:18for her two directed therapies.
01:20I think this is particularly timely as
01:23the first her two targeted therapy for non.
01:26Fish amplified breast cancers was
01:28just approved six months ago and how
01:31exactly we identify which patients
01:32are going to benefit from this therapy
01:35I think is a huge question and the
01:37one the field is struggling with and
01:39David's made a lot of inroads into that
01:41and I think he's going to focus on,
01:42on on that today.
01:44So thank you for bringing this
01:46timely discussion to us.
01:51OK, great. Thanks, Diane, and thanks to
01:54the leadership for inviting me today.
01:56But thanks especially for
01:57Ian for introducing me.
01:58He's a world leader in this space
01:59that I'm going to talk about.
02:00That is the her two.
02:0480C or antibody drug conjugate space.
02:07I'll start by my title is precision
02:09medicine versus persuasion medicine and
02:11I'll get to what persuasion medicine
02:13is a little bit more at the end.
02:15And reading versus measuring.
02:17And measuring is what you do quantitatively.
02:20Reading is what pathologists do
02:22when they look at slides and
02:24difference between subjective and
02:25objective assessment of tissue.
02:30Let's see. Let's start with my disclosures.
02:33As you can see, I do a fair bit of consulting
02:36and a lot of the research in my lab,
02:38including the work that led to this.
02:40Most of what I'm going to present
02:42today was sponsored by companies
02:43including Sephylon and Kanika Minolta.
02:47So today I want to spend the next 55
02:50minutes or so talking about first and
02:52quick introduction to the new drugs.
02:54If you've heard Ian speak,
02:55I don't need to give this part,
02:56but maybe you didn't a proposed new assay
02:59for these new drugs high sensitivity
03:01would call the assay high sensitivity
03:04or HS or two and then CAP CLIA,
03:06what take, what does it take to get
03:08something from your research lab so
03:10that into a lab where we can deliver
03:12information to patients and put the
03:14results in the patient's chart.
03:15That's what cap CLIA is taking the new.
03:18Say to the clinic and then finally I'll
03:20talk about precision medicine versus
03:21persuasion medicine and try to talk all
03:23of you who are oncologists in the room
03:25and to not using persuasion medicine
03:27and focusing on precision medicine.
03:31So this is the, the big drug that got
03:33the first standing ovation in 25 years
03:35as I understand at ASCO and it it's,
03:38it's the same old drug,
03:39it's trastuzumab underneath,
03:41but they've conjugated 8 topoisomerase
03:44inhibitor payloads to the trastuzumab that
03:47gives you especially some special tricks.
03:50First of all, it brings these highly
03:52toxic payloads right to the cell.
03:53So it doesn't have the toxicity that
03:56giving the drug and the dosages that
03:58it would be given cause would cause.
04:00All kinds of toxicity.
04:02But if you bring it right to the
04:03salad it causes less toxicity.
04:05Whereas if you, and not only that,
04:07when it gets uncoupled in the
04:09cell it can spill out of the cell
04:11and kill neighboring cells.
04:13The so the the sort of neighborhood effect
04:16or proximity effect of the therapy.
04:20It worked really well and that's
04:21why we've all heard about it that
04:23you can see very few patients were
04:26resistant but most patients had
04:27some response and there were eleven
04:29CR's in the early trials and in
04:32fact it worked at for all patients,
04:35but especially in patients that
04:36were not amplified for her too.
04:38So the initial trials were all in
04:40patients that had her two amplification,
04:42but then they started trials on
04:44patients with low her two IHC 2
04:46plus and HC1 plus and you can see
04:48the curves look pretty similar.
04:50And in fact in those low,
04:51low patients in the Destiny 4 trial,
04:54ultimately the survival curve
04:55looks like great, looks like this,
04:57which is really a great improvement
04:59in the survival curve for advanced
05:01breast cancer and changing median
05:02survival from 5 to 9 months.
05:04And that's I think what ultimately
05:07has led to the popularization of this
05:09drug and the success of the drug.
05:11And they said we concluded a
05:13randomized 2 group open label phase
05:15two trial with her too low.
05:17What does that mean?
05:18So that's what we'll examine the rest, but.
05:20Before we go there,
05:21what about her 20?
05:22What about if they don't express
05:24any her two at all and can we tell
05:26the difference between her 20 and
05:27her two low and in fact in her two
05:30zeros and this there is a trial
05:31going underway that's her two less
05:33than one but greater than zero.
05:35That's the Destiny 6 trial
05:37hasn't reported yet.
05:38But there's also the her 20 equal 0
05:40Daisy trial which was a small trial
05:42in France where there was clearly
05:44in these waterfall plots clearly
05:46patients that benefited from drug
05:48even though they had a hurt 2 = 0.
05:51So is this the why is it important
05:53to understand this and have good
05:55diagnostics for it because this
05:56drug is the tip of the iceberg.
05:58Here's a list of other drugs
05:59which there's no way you can read,
06:01but all of these drugs are,
06:02are all these are targets for
06:04ADC's in clinical trials.
06:06So I think ADC's may become very
06:08important for oncology in the next
06:10few years and equally important
06:12will be companion diagnostics that
06:13actually pick the right patients as
06:15opposed to giving the drug because
06:17unlike when we know the tiger it so
06:19well and we know how the drug works.
06:21It's really important to be able to
06:22pick the right target or to pick
06:24the right patients that express
06:25the right amount of target.
06:26So what do we do now?
06:27So this is the standard practice guidelines,
06:29ASCO CAP guidelines in 2018 and
06:32these guidelines are how we practice
06:35as pathologists in assessment
06:37of her two expression.
06:38And this is the algorithm for what
06:41we look at when we look at the slide
06:43circumferential staining that is complete,
06:45intense and greater than 10% of the cells.
06:48That makes the three plus
06:49and then we have a 2 + 1 +.
06:50I won't go through them all,
06:52but they're kind of summarized
06:53here where one no staining,
06:55no membrane staining observed is a 0 +
06:571 is faint partial membrane staining
07:00and weak to moderate staining is +2.
07:02That's kind of subjective. In fact.
07:05How well can we do that and
07:07how important is it?
07:07Well.
07:08It used to be important to tell
07:10the heart threes the the three
07:12pluses from the others and the
07:14two pluses were they the reflex.
07:16But now it's important to have this CAD.
07:18The new category is far too low.
07:20And how many are there?
07:21There's a lot,
07:22maybe as many as 65 or 70% of the
07:25patients are thought to fall into this low,
07:27her two low category which means
07:28a lot of patients could get drug,
07:30but it also means that we need to be
07:32as accurate as we can and assigning
07:34those patients because we don't want
07:36the her two zeros to get the drug
07:38if they aren't going to benefit.
07:39Well, maybe we do.
07:40We'll talk about that later.
07:41But so how do we know how,
07:43how well do we do at this?
07:44Her too.
07:45So I'm fortunate to be on the
07:47immunohistochemistry committee of
07:48the College of American Pathologists.
07:50And so I get access to the
07:52surveys that make Clea labs,
07:54what CLIA labs are.
07:55That is,
07:55for a CLIA lab or clinical lab
07:57to return data to the chart,
07:58they have to do a survey twice a
08:00year to show that they're competent
08:02and effective at doing the assay.
08:04And here's the surveys for anatomic
08:06pathology for her two using a tissue
08:09microarray. This is her to a 2020.
08:11So it was the fall survey or the
08:13spring survey from 2020 from the
08:15College of American Pilot Pathologists.
08:17And you can see my colleagues here,
08:18including myself, who are on this committee.
08:20And when we looked at these surveys,
08:22we noticed that four, six and seven,
08:24that is 3 out of 10 did not reach consensus.
08:27That means that of the 1400 labs
08:29in the in the world that did this,
08:32they couldn't come to an agreement.
08:34That is,
08:34you need to have 90%
08:36consensus to have agreement.
08:37In fact, if we look at this one,
08:39it's interesting.
08:39This is one of the cases that didn't
08:42come to agreement and that was
08:44because there was a big discordance
08:45in the number of called 0 versus
08:471 and there were a fair number
08:49that even called it two or three.
08:50So that's troublesome.
08:51If we're testing these labs twice a year
08:54and we're assuring that they're giving
08:56the right answer for the patients,
08:58how can you have that much difference
09:00between zero and one that it's almost 5050.
09:03So since I I'm on the CAP committee,
09:05I could ask for the data from
09:07the last few years.
09:08And here's the data from
09:09the lab from 2019 and 2020.
09:12And of the 80 cases,
09:14fifteen of those cases showed a
09:16discordance of greater than 25%.
09:17And that's shown in these pie
09:19charts here where the zeros are
09:21blue and the ones are red and
09:23anything higher than one is black.
09:25Two and three,
09:26since we're not going to focus on that.
09:27So we we did,
09:28we thought is this really, you know,
09:30this is concerning,
09:31but you know what this is tissue microarrays.
09:33This is not what happens in the real world.
09:35So then we did a study of real
09:37world core biopsies.
09:38And enrolled 18 pathologists from 15
09:40institutions around the United States
09:42and ask them to read actual core
09:44biopsies that have been read at Yale.
09:46We collected 170 cases from Yale
09:48and had them score them according to
09:51the ASCO CAP guidelines before the
09:54publication and the popularization
09:55of her two one plus versus 0.
09:58So they didn't know they were just
10:00doing the ASCO CAP guidelines as
10:01they always have, and scoring 0123.
10:04And what they did,
10:05what they're scoring looked like was this.
10:07That is, the Blues were the zeros.
10:08This is the percent of pathologists.
10:10That scored at 0,
10:12so a fair number agreed that
10:14that there were zeros.
10:15There were 92 cases that were discordant,
10:18and of those,
10:209269 were discordant between zero and one,
10:23and only twenty were discordant
10:25between 2:00 and 3:00.
10:26So this actually was.
10:28Through many reviewers,
10:30ultimately got us published in JAMA Oncology.
10:34How come it's not advancing now?
10:38Oh, hold on. I lost my laser pointer.
10:44Microsoft doesn't want me to do this now.
10:45It's the screen has turned
10:47Gray as it's saying restart.
10:49I probably shouldn't.
10:50I should probably wait for
10:51the program to respond.
10:53Very sorry about this,
10:54but suffice it to say that I'll skip
10:57the next slide so we can keep going.
10:59The next slide was after
11:00many review rounds of review,
11:02we did get this published in JAMA Oncology,
11:04but weren't allowed to say
11:05what we wanted to say,
11:06which is that there's really a
11:08great discordance between zero and
11:09one and not so much discordance
11:11between 2:00 and 3:00.
11:12And I don't know if we have
11:13should I restart the program?
11:14I don't know if we have any IT
11:16people here that or how long we're,
11:17you know,
11:18we'll be here for the next 45 minutes
11:20waiting for the computer to come along.
11:25Wait to respond or restart the program.
11:28Maybe I should restart the program and
11:29that will take several minutes too.
11:31It's just it's just not rebooting
11:33the computer, it's just. Yeah.
11:37Let's try again. Yeah, we're good.
11:41We're good. Sorry about that. OK.
11:43We saw all those things already.
11:45Let's go to. The study that was in
11:49JAMA Oncology was here and this is
11:52what was the this is the figure and
11:54and Eileen Fernandez was the lead
11:56on this study in my lab and she.
11:59Did the analysis that is shown here
12:00that shows that there's a lot more
12:02discordance between zero and one
12:04than there is between 2:00 and 3:00.
12:06And for two we have a solution.
12:08For two we can do fish,
12:09so we have a orthogonal assay.
12:11What do we do between zero and one?
12:12Well, we don't have a solution yet.
12:14That's what I'll show you in a minute.
12:16But also you can look at this analysis
12:19which which shows you this is a work
12:21done by Jack Robbins in the lab with
12:24Eileen Fernandez showing the percentage
12:26of people that called 0 versus called one.
12:29And so if your pathologist #18 and
12:31these are all currently signing out
12:33pathologists most with more than five
12:35years of experience around the country.
12:37So these are not residents or not to say
12:39that residents can't do this just as well.
12:42But these are not residents
12:43or or or laboratorians.
12:45These are sign up with ologists.
12:47And if you're pathologist 18 you only
12:49score 15% of the patients with a 0,
12:52but if your pathologist number one,
12:53you have 44%.
12:54So whether or not you get trustors,
12:57mab drugs,
12:57tecan depends on who your pathologist is.
12:59That doesn't sound like a great idea to me.
13:02So what we asked is how many people
13:04do you need to make sure that
13:06an assay agrees with each other?
13:08And we, we invented this with gang hand.
13:10We invented this system to realize that
13:13there's 21,000 pathologists just in the
13:14US and at that and 100,000 in the world.
13:17So how many do we need to decide
13:19whether or not an essay is good?
13:20And so we there, there is actually
13:23no statistical method for this.
13:24So we simply decided to plot
13:26the overall percent agreement.
13:28That's what overall LPA stands for
13:30versus the observers or readers.
13:32And what you see is that the more
13:35observers you have,
13:36the less agreement you have,
13:37which makes sense.
13:38The more people you ask,
13:39the more discordance you're
13:40going to get in your answers,
13:41just mathematical truism.
13:42And so does this actually work and
13:45can this be used to assess assays?
13:47So here's estrogen receptor.
13:48Turns out we're really good
13:50at estrogen receptor.
13:51If you have a quartet of
13:54pathologists read estrogen receptor,
13:56all four of them will agree somewhere
13:58between 85 and 95% of the time.
14:01How do we do for her?
14:02Too well, not so well.
14:05This is the plot for her.
14:062/3 plus or not three plus.
14:08We're really good at that.
14:09But if you have a quartet of pathologists
14:124 decide whether it's zero or not zero,
14:15it's between 40 and 80 percent,
14:1885%. So this is a new method
14:20to approach the analysis,
14:21to try to figure out how many we
14:23need and how many do we need to make
14:24a new assay to make a good study.
14:26Well, it's when it plateaus,
14:27so in this case we probably need 9 or 10.
14:30And this case, no number is sufficient
14:31because it goes all the way down to the
14:34baseline to tell ones from, not once.
14:36So the point is, I think that.
14:38I've I hope that I've convinced
14:40you that we need a new assay.
14:43And so that's what we have done,
14:44is propose a new assay that's measured,
14:47not red. And so based on that.
14:50We start, we started from the beginning
14:52with cell lines and these cell lines
14:54are all cell lines that are amplified,
14:56gene amplified and these cell
14:58lines are all gene express.
15:00Her too,
15:01but are not gene amplified and
15:03you can see when you plot them.
15:05If you look with the current FDA
15:07approved assay you can separate the
15:09highs from the lows or the negatives,
15:11but you can't stratify the negatives.
15:14Whereas if you do the new assay
15:15high cut 10 times more antibody,
15:17pretty simple new assay.
15:18You can then stratify the low cell
15:20lines and tell the zeros from the ones.
15:23Essentially if you were reading the
15:25cell lines but it was the wrong,
15:27the wrong the current assay is
15:29the wrong tool for the job.
15:31Or as was said by a group in France.
15:36The current assay now FDA approved,
15:38is like weighing mice on a
15:40scale for elephants.
15:41And I think this is really good
15:43because everybody gets this if
15:44you have a skill for elephants,
15:45it doesn't work for weighing mice
15:48and it's all about dynamic range.
15:50So here's the assay we did we
15:52invented and and this is to have
15:54a a series of cell lines and then
15:56just do like a Bradford assay like
15:58we all did in college chemistry
16:00for where we make a standard curve.
16:02And we used our tissue microarray to
16:04make cell lines and with the help of
16:06array science made a standard curve.
16:08And then with the help of Crotia we
16:10figured out how many animals per
16:12microgram there were in each of these
16:14cell lines and then converted that
16:16using Q path to how many animals
16:18per square millimeter there are.
16:19So now we have an assay.
16:21That can tell us animals per
16:22square millimeter.
16:23And like all assays,
16:24it saturates when it gets too high.
16:26So the amplified cases are
16:28saturated and we can't use those.
16:30But since we don't really care
16:31about 2 plus and three plus,
16:32we got that pathologists can do just fine
16:34in telling 3 plus from not three plus.
16:36We need an assay to tell 0 from 1.
16:39And so that's this assay works fine.
16:40If we get rid of these two,
16:42we can now build a very nice
16:43standard curve that we can use
16:45as a linear assay and then assign
16:47each case and animals per square millimeter.
16:49And so just to remind you.
16:51I'm going to talk a little bit
16:52about limits of detection,
16:53limits of quantification,
16:54and limits on linearity.
16:55A little bit of essay terminology and we
16:57and this is the range we want to be in,
17:00not this range,
17:00which is what the saturation range,
17:02which is what the current
17:04essay really focuses on.
17:06Because really the current
17:07assay all you need to tell is.
17:09Is it saturated or not?
17:11For the new assay we need
17:12to tell how much they have.
17:14So here's our the current our standard curve.
17:17With the higher two assay and
17:19you can see there's two positives
17:20and the rest are negative.
17:22So it works if you just want to
17:24tell amplified from non amplified,
17:26but what if you want to tell that
17:27low range so you can see that you
17:29can see the full dynamic range with
17:31the new her two low assay antibody
17:32concentration or that what we're
17:34calling high sensitivity HSV or
17:36two you can see in that range.
17:38So now we have to talk about a
17:39little bit of wonky stuff and
17:41that is what are those things.
17:42So what is the limit of detection,
17:44what is the limit of quantification
17:45or what is the limit of linearity.
17:47So these are the definitions
17:48and this is right.
17:49One of the FDA's handbook on how they
17:51advise industry to do this and you
17:53can see that the limit of detection
17:55is the lowest concentration of the
17:57analyte that can be detected but
17:59not but and reliably to strings
18:01from zero but not necessarily
18:03quantified that is too low.
18:05So what we really want is to know
18:07the limit of quantification because
18:08then we can do it right every time.
18:10But we don't yet know how much
18:12her two is required to benefit
18:14from trust who's mab drug stcan.
18:16So we're going to measure all the
18:18way down to beyond the limits of.
18:20Our essay to to the LD and below and
18:22see what we get and what we got is this.
18:25When we did it on tissue microarray we
18:27could see that the the zeros are blue,
18:30the Reds are one, ones are red, the twos.
18:32This is the pathologist read over here
18:342 Plus is black and three plus is green,
18:36and most of the Greens are
18:37above our limit of linearity.
18:39But look at how many twos and ones
18:41there are in this middle range.
18:43That would be called one.
18:44And I think this is even further evidence
18:46that we need a a quantitative assay.
18:48We need a measured assay,
18:49not a red.
18:50Say,
18:50in order to pick the right
18:51patients for trastuzumab drugs,
18:53tican and surprisingly there are some
18:55patients most of whom were called 0,
18:57but some were called one or two
18:59that are actually below our limit
19:01of quantification or even below our
19:02limit of detection as I'll show later.
19:04So then we did what you have to do in
19:06a clear lab is did 40,
19:08you have to do 20 positives and 20 negatives
19:10according to Fitzgibbons at all in order
19:12to bring your assay to the CLIA lab.
19:14But we don't have positives and negatives.
19:16We have a continuous scale.
19:17So we did 40 of them and these
19:19are actual core biopsies.
19:20Now they're not tissue microarrays,
19:22but you can see the same thing.
19:23There's a fair bit of Miss Assignment
19:25and in fact summarized here,
19:27you can see that there's zeros and ones,
19:30but there's a broad range of animals per
19:32square millimeter for zeros and ones.
19:34And the two plus not amplified almost
19:37in fact does overlap with the two
19:40plus amplifies and the three pluses,
19:42which we're good at and we're pretty tight.
19:45So how many are there?
19:46Well, in our first forty there
19:48was about 20% of the cases that
19:50appear to be below the limit of
19:52quantification for her two protein,
19:54but potentially present and as
19:57target for a target for TDXD.
20:01So just to summarize to this point,
20:03about 70% of the cases have low her two
20:06defined as above the LQ and below the
20:09levels associated with gene amplification.
20:11About 8 to 10% are below
20:13our LQ or even our LD.
20:15It's probably about 6% below our LD.
20:18Many of the cases that
20:19are called her to zero,
20:20as many as 60% are in our studies,
20:21maybe 75% have detectable amounts
20:23of her too between 3 and 20 animals
20:26and the quantitative her two asset
20:29could be envisioned as a reflex tax.
20:31So that if you pathologist reads
20:33an IHC equals zero,
20:35they could then reflex to the
20:36quantitative test and the same way we
20:39reflex to fish today for A2 plus HC.
20:43OK, so that's the proposed new assay.
20:45Now let's take it to the clinic.
20:47So what's involved in the next
20:48step of taking to the clinic?
20:50And I like to quote a colleague
20:51of mine from Brigham and Women's
20:53who said once you have the essay
20:55working in your research lab,
20:56you're 5% of the way there.
20:58And I think that's really true.
20:59Now having brought this assay
21:01with hats off to Trish Gal who's
21:03not here and has not left,
21:05Nay Chan who was in the audience and Reva
21:08come ova who have helped me to bring
21:10this assay to the clinical setting.
21:12So the things that you have
21:14to do are antibody titration,
21:15maximization of signal to
21:17noise analytic validation.
21:18I'll try to go through this stuff fast
21:19because it's a little on the wonky side,
21:21performance accuracy, precision,
21:23sensitivity and specificity and
21:25serial core reproducibility.
21:26And then how do we tell our colleagues?
21:28What do we tell the oncologists?
21:29And then so the reporting is
21:31part of this as well.
21:32So first of all,
21:33we looked at the signal to noise
21:34and you can see that the peak
21:36signal to noise is at 1 microgram
21:37per mil for a new antibody.
21:38This is a new higher
21:40sensitivity antibody for her
21:41too than the one that's
21:43currently used in the clinic.
21:45And we took and we picked the
21:46concentration with the maximal signal
21:47to noise and then we looked at the
21:49accuracy and our accuracy isn't great,
21:51it's only 87%. Why is that?
21:54That's because we're more sensitive than
21:56the status quo assay which we had to
21:59compare it to which was HC012 and three.
22:02But overall we have quite a quite
22:05good concordance especially in the end
22:07and more resolution in the low range.
22:09And then our intra and intra
22:11assay precision is quite high,
22:1310% sounds like it might not be great.
22:15To interact assay precision and actually
22:17the essay that we just bridged to,
22:19we're now under 10%,
22:20but it's acceptable and the
22:22intra assay precision,
22:24this means to calculate the precision
22:26three slides run on separate
22:27trays at the same machine is,
22:29is, is about 5%.
22:30So these are where we want to be.
22:33Our sensitivity compared to the historical
22:35essay as 100% and our specificity is 84%.
22:39Why is our specificity low?
22:40Because we're more sensitive and
22:42so we call cases positive that
22:45we're called negative by IHC.
22:47So here's the proposed clinical
22:49future work workflow and this is
22:51what we're doing now which is we
22:53have we get the labs come to this
22:55lab that I've called the qutab lab
22:57quantitative diagnostics and anatomic
22:58pathology which is a new lab which
23:00is now open and open for business.
23:02And we've now begun to do this.
23:04This is and this is qdap essay #1,
23:07the high sensitivity here.
23:08Two,
23:08we batched the stains and do them
23:10in our like a bond stainer so that
23:12they're done in an auto stainer
23:13and then we read them originally
23:15in some old like legacy hardware.
23:17But now we're using this,
23:18we just recently completed the bridge study,
23:21although our license holder
23:22hasn't signed off yet,
23:23he will see it shortly and uses a
23:26much more high throughput device.
23:28Instead of an hour this machine would
23:29take about four minutes to scan a slide.
23:32So we wanted to update our technology
23:34a little bit and then we signed
23:35it out and Co path as a procedure.
23:37And so it ultimately makes it to
23:39epic and and clinicians can see it,
23:41this is what it looks like.
23:43The pathologist has to pick a region.
23:44So we're actually not measuring
23:46the entire core biopsy.
23:47We're measuring a region that is
23:49quote UN quote representative and
23:51that representative region is
23:52then looks like this.
23:53This is actually not a brown stain
23:55but a pseudo IHC which it shows the
23:57pathologist what they what it looked
23:59like and then the pathologist actually
24:01sees the number of fields of view,
24:03in this case 23 and the in this case
24:06the rare sight score in this case was
24:0915.4 animals per square millimeter.
24:11So that will be included in the report.
24:13We'd say 15.4 animals require millimeter.
24:15We don't know what that means.
24:17I mean,
24:18we do know that it's detectable
24:19and then
24:20we can give a choice in our interpretation
24:22that it's positive for expression high.
24:24That is, it's above our limit of linearity
24:26positive expression intermediate,
24:28which means that it's like a one
24:30or A2 positive for expression low,
24:32which means it's.
24:33Present, but it might not be reproducible.
24:37That is, it's above our LOD but
24:39not necessarily above our LOQ
24:41and then negative below the LOD.
24:43And so these are the reports that we'll
24:45issue as as we start to receive specimens.
24:48So far we've received a grand total of two.
24:51We hope that after this talk and maybe
24:54in the future and certainly in the more
24:57distant future when we know how much.
24:59Is necessary for patients to respond.
25:02We hope that this essay will
25:03gain some traction.
25:04So our vision we currently offer HSR 2 in
25:07the QDAP lab test must be requested by an
25:10oncologist and the patients are billed.
25:11If the test is requested by an oncologist,
25:14there are I CD9 codes for all
25:16the stuff we're doing.
25:18We began a prospective study on all
25:20breast biopsies so that we have data of
25:23a year's worth of prospective data and
25:26we're about seven months into it now.
25:28We offer the essay coalitions
25:30from to Yale or elsewhere who want
25:33quantitative information,
25:34but only two so far to date.
25:36And then the discussions of the
25:38license we will.
25:38What we hope to happen is ultimately
25:40it won't just be yell that can do this,
25:42but we'll license it to some of
25:43the big lab companies that provide
25:45them the bulk of the service.
25:47It's interesting to know and
25:48interesting to me anyway,
25:49that only 15% of lab tests in the
25:52US are provided by academic labs.
25:54The other 85% are provided by private labs.
25:58And so clearly if we want to.
25:59Have this effect patients around
26:02the world and be useful and needs
26:05to make it into private labs and
26:07those discussions are beginning.
26:09So the last thing I want to talk about is
26:12the precision versus persuasion medicine.
26:15And so our original envision for
26:17this essay was that we would need
26:20to adjudicate the IHC's equal 0.
26:21And what we would do is we would get
26:24all the HC equals zero and we would
26:26measure them and then we tell you if
26:28you're above the limit of detection
26:30or above the limit of response.
26:31We don't know the limit of response yet.
26:33Someday we will and I'll show you
26:35how we intend to get there.
26:36But right now we don't know the
26:38limit of response but.
26:38You would take all the cases that
26:40were called HC0 and maybe the cases
26:42that were called HC One and do that.
26:45But something happened in the last
26:46three or four months and I haven't
26:48been able to document it yet,
26:50probably because it's not mature enough,
26:52but suddenly the IHC equals zero is rare.
26:55And that's because pathologists
26:57are people too.
26:58Pathologists sometimes might be
27:00a little more lenient on what
27:02they call IHC one and this code
27:04called Sympathy vote because
27:05then they can get this new drug.
27:07Here are real quotes that I've heard.
27:09I won't quote the people because to
27:12not embarrass them or give them credit,
27:14but here's a real quote.
27:16Hi doctor pathologist.
27:16So I see you called Missus
27:18X's biopsy IHC zero.
27:19That means I'm going to have
27:21to offer her brain radiation.
27:22Are you sure it's not H sequels one?
27:24Then I could give her her and her two.
27:28Should I go look at that slide again?
27:30Does that mean that my first view
27:32of that slide was not accurate?
27:34Or was it accurate and maybe
27:36I should change my diagnosis?
27:39Because I'm persuaded that
27:41that's better for the patient.
27:43I'm not sure that's a great idea from
27:45West Coast director of pathology service.
27:47Yeah, we don't have many
27:48IHC's equal 0 anymore.
27:50And from a Midwestern oncologist,
27:52I'm not seeing the response rates
27:53and in her two patients that
27:55they saw in the clinical trial.
27:56They're getting a lot of IHC zeros and
27:59maybe IHC zeros really don't respond.
28:01We know that eight to 10% of the
28:04cases really don't express any target
28:06and this is a targeted therapy.
28:08I mean we don't definitively
28:09know how the drug works,
28:10but we think it's a targeted therapy.
28:13After all,
28:14it's trastuzumab conjugated to toxins.
28:16So what's happened is that really
28:18now we need to adjudicate the one
28:21pluses what we really need because
28:23the zeros have minimized, not,
28:25I don't want to say they've gone away.
28:27If you ask pathologists,
28:28they will sternly tell you yes,
28:30of course we still call IHC 0.
28:32But.
28:34Data will set will will tell us
28:36in a year or so from now how
28:38our IC0 calls changed.
28:40But but I see one is now more common
28:42and so if it's if it's more common
28:44maybe that's the one we should be
28:46measuring and in fact that's the plan.
28:48So there are a few different ways
28:51we're going to study IHC equals one.
28:53The first is the Qdap Labs
28:56prospective study and this is
28:58copied with me by name by Nate Chan,
29:01who's the director of the Q dot lab.
29:03And you can see we began August
29:051 and we'll go till July 2023
29:07and today we have 226.
29:09I anticipate we'll get around 400.
29:12The inclusion criteria will be any
29:13case and the primary objective will
29:15be to determine the number of H0 cases
29:18that have detectable her two expression.
29:20So how many IHC zeros?
29:21And this study was designed
29:23before everything.
29:23Name HC One,
29:24but how many HC Zeros have above
29:26the limit of detection and how many
29:28HC ones have below the limit of
29:31detection will be interesting as well.
29:33That's a secondary,
29:34that's a secondary objective of the
29:36study and the study is in process
29:38and we all just to show you a peak,
29:41we've already started doing some
29:42quantitative work and in fact
29:44you can see from quantitative,
29:45this is quantification of prospective
29:48tissue from the clinical trial
29:50and you can see the lol in this
29:53case was 33 and the OD is 3.
29:55This is all done on the new platform
29:57and you can see that there's a lot
29:59of cases that are called zeros that
30:01are above our limit of detection.
30:02There's not as many.
30:03So far it looks like we're going
30:05to not have very many that are
30:06below our limit of detection,
30:08but time will tell as we get
30:10as the study
30:10matures. There's two other studies
30:13that were progressing on one is a
30:16TB CRC study report proposal with
30:18Ian and Eric's arrival at Yale,
30:20we became part of the Translational
30:23Breast Cancer Research Consortium,
30:24which is a group of 16 or 17.
30:26Now institutions that do studies
30:29together on translational research
30:30and the goal of this study that
30:33is still in proposal stage is to
30:35evaluate her two measurement in
30:36the one plus metastatic cases.
30:38So if we get one pluses and we get 2 or
30:41300 from 17 institutions around the country,
30:44we should be able to tell how
30:46frequently we see the patients
30:48that have one plus actually don't
30:50have any target and vice versa,
30:52we should be able to see response
30:54since all those patients since they
30:55were called one plus will be get.
30:57Drug will be getting trustors map drugs
30:58he can and be present in the residency.
31:00So here's the study a draft of the
31:03study objectives to evaluate the
31:04real world relationship between
31:06quantitative her two expression QIF
31:08and objective response in patients
31:10with her two IHC plus one and
31:13metastatic breast cancer receiving TXT.
31:15And then there's a number of secondary
31:17objectives that are shown here as well.
31:19And then a second study that I,
31:21I don't even have a slide for yet
31:24is that we proposed a study led by
31:26Merriam Lustberg here of patients
31:29who get HC0 and then prospectively
31:31giving them TXD much the way the
31:35Daisy trial worked on that study
31:38is not yet completely designed
31:40and not yet completely approved.
31:42So I don't have any slides to discuss it,
31:44but I think those are the kinds of studies
31:46we need where we have patient response.
31:48Either real-world patient response or
31:50clinical trial patient response in order
31:52to figure out the animals per square
31:54millimeter above which patients benefit.
31:56Will it be a cut point?
31:57Probably not.
31:58Probably there will be patients with
32:00high animals per square millimeter
32:02that still don't benefit because there
32:03are other mechanisms of resistance.
32:05And I have and one of the interesting
32:07topics that many labs are working
32:09on including my own are what are
32:11the mechanisms of resistance beyond
32:12just not enough her to present.
32:14And hopefully next year or a couple
32:16years from now I'll come back to you at
32:18grand Rounds and talk about mechanisms
32:20of resistance and a more complex assay
32:22that also doesn't just assay her too,
32:25but maybe assays other biomarkers
32:27that are associated with resistance.
32:29Or other drugs.
32:30And in fact the her two trope
32:312 assay as well along its way.
32:33So we can help clinicians decide
32:36between Saskatoon Vova Tican,
32:37which is a trope 2 targeting therapy
32:41versus trastuzumab Drexel can.
32:43So for that my my last slide,
32:45overall HSR 2 assay is an LDT,
32:47a lab developed test and not FDA approved.
32:50So if you only do FDA approved tests,
32:52you probably don't do them here
32:53since most of our assays are LDT's,
32:55but we do have a few FDA approved assays
32:59and many FDA approved assays.
33:01People don't realize this,
33:02but being on the CAP
33:03committees you realize this,
33:04if you change one step of the
33:06protocol of your FDA approved assay,
33:08it is then an LDT and you must thus
33:11validate it and so most assays.
33:13We do are not FDA approved.
33:15We might use FDA approved reagents,
33:17but most assays we do are actually LDT's in
33:19our lab and in all the labs around the world.
33:22And that also applies for molecular assays,
33:26gene mutation assays.
33:27Many of those assays are also not
33:29FDA approved assays but rather LDT's.
33:32HSR 2 essay is in the correct dynamic range.
33:35That is, we're not weighing elephants on or
33:37weighing mice on a scale built for elephants.
33:39The level of target required for trustees,
33:41mab drugs decan is still unknown and
33:43I speak here before you and I don't
33:45want to try to hide that from you.
33:46I think it's very clear that we don't
33:48know the answer to this question yet.
33:50But if we waited until we knew the answer to
33:52the question before we started the essay,
33:54we would be years behind as as this essay.
33:56We've been working on this essay
33:58for a couple years now to get it
34:00to the point that it's at.
34:01And so now that we have.
34:03Tools, I am asked that oncologists in
34:05the audience ask for measurements,
34:07not for readings.
34:08And please don't ask the pathologist
34:10to change their minds.
34:12That's persuasion Madison,
34:13not precision medicine.
34:15And we all respect our pathology
34:16colleagues and I think we all,
34:18you know,
34:19I know that oncologists really think highly
34:21of most of the pathologists they work with.
34:23And I think that they don't realize
34:25that when they do pursue persuasion
34:28medicine that it's actually not
34:29what the biologist wants to hear.
34:31They don't want to be second guessed.
34:33They want to if if if we're giving you
34:35a reading, we're giving you a reading,
34:37we really believe that's right.
34:39And just like you shouldn't go back
34:40on the test and change your answer,
34:42don't change your answer.
34:44It's if.
34:45If that was your first impression,
34:46it's probably your true impression
34:48and probably your best reading.
34:50And so with that,
34:51I just want to thank the people
34:52in lab that do all the work.
34:53I get to talk about it,
34:54but it's really a crew of people that do
34:56all this stuff that I told you about.
34:58I especially like to point out mirror
35:00to Matafi who started that and started
35:02building this essay in the lab over
35:05two years ago now and then my Yale
35:07collaborators and funding sources etcetera.
35:09And then here's the the the key
35:12group at our last holiday party,
35:14our lab group Aileen has now left.
35:16She was involved in a lot
35:18of the analytic stuff.
35:19Matt Lou helped out with some of the.
35:20Analytic stuff as well.
35:22And then and Jack and Katie
35:24weren't at the party,
35:25so they got their picture separate.
35:27So with now, I've also left
35:28about 20 minutes for questions,
35:30if there are questions.
35:31Thank you very much.
35:40So we have 4 questions in the chat.
35:43Maybe while you're warming up,
35:44should I start with those?
35:45Oh no, there's only two.
35:46What about discordance with
35:48pathologists reading the same
35:49slides after a washout period?
35:51So Manju Prasad,
35:53a esteemed pathologist in our department,
35:55asks a very pivotal question, that is.
35:58When you're doing any kind
36:00of pathologist study,
36:01when you read it once,
36:02if you're going to read it again,
36:03you should have a washout period.
36:04That is so you don't remember
36:06that case because surprisingly,
36:07pathologists have a really good memory for
36:09what the morphology of cases look like,
36:11and they can also remember the
36:13patient's name on the label.
36:14And so a lot of studies have
36:16a washout period.
36:17We didn't need a washout period in this
36:19study because they only saw the slides once.
36:21So if we're going to show them to them
36:23again and if we're going to do any
36:25kind of intra observer reproducibility,
36:27which we didn't do and some
36:29other studies have done,
36:30we would need a washout period.
36:31But in this case,
36:32a washout period was not required.
36:34And then Timothy Robinson asks,
36:37is heterogeneity within
36:38the tumor important issue?
36:39Is it more important to do a small
36:41percentage of cancer cells that
36:42express a high amount of heart, too?
36:43Or is it more important to know that
36:45a high number of cells expressed at
36:47least the minimum amount of her too?
36:48Wow, phenomenal question.
36:50That's Jax three, that's his.
36:51Thesis project,
36:52I think that's a great question.
36:54We obviously don't know the answer.
36:56All the pathologists in the audience
36:57know that her two is very heterogeneous.
36:59Not only is it heterogeneous
37:01from within a slide,
37:03but it's heterogeneous between cuts.
37:05And all the pathologists in the audience
37:06know that when we sample one core
37:08biopsy that's less than 1% of the tumor.
37:10And so there's no way for us to actually
37:12answer that question about true
37:14heterogeneity of the patients tumor.
37:16But what we can add,
37:17we can ask about heterogeneity
37:18on the slide and we can and are
37:21asking at that question.
37:22That is,
37:22how important is high expression
37:24in a single cell versus high
37:26expression in the average cell?
37:28We started with the average.
37:29You have to start somewhere,
37:30and I don't know that the
37:31average is a correct answer.
37:33You could argue because of
37:34the bystander effect of TDXD,
37:36it's actually the highest ones
37:38that make the most difference,
37:40but we don't know that.
37:41That's just speculation at this point.
37:45Let's see. Now it's your turn.
37:49It doesn't.
37:51Say.
37:55That's a great question.
37:59We can do more.
38:00So as soon as I had the assay built,
38:02I applied for tissue from AstraZeneca
38:05adicci senko from the Destiny 4 trial
38:08and was rapidly told I would never see that.
38:11And it's, I don't fault them for that.
38:14They have their own people that
38:16can do quantitative work and they
38:19have an FDA approval for IHC 0123.
38:21So they don't want to have to
38:23change their FDA approval.
38:24They're making a lot of money on this
38:26drug and it would be detrimental to
38:28the shareholders of that company to
38:29have me have access to that tissue.
38:34My question was how much heterogeneity
38:37do you see in the ATOMAL expression?
38:39Because you're taking so many fields of
38:41view and taking an average, do you see
38:43a lot of heterogeneity there or is it?
38:46So that's a great question.
38:49Heterogeneity within a core
38:51biopsy is quite substantial.
38:53And as you know when you read them,
38:55you see bright areas and not so bright areas.
38:57And you know how do we handle that?
38:59Well, someday we'll know how you
39:01know whether it's the highest sell or
39:03the average sell or the lowest sell.
39:06That's most important for response
39:07to Trump drug seeking.
39:09But we don't know that yet.
39:10And so in the same vein of OK,
39:13we're just going to take a core biopsy and
39:15say that that represents the whole tumor,
39:17we're going to just take the
39:18average and say that that.
39:19Represents the expression of her too.
39:22And the second question
39:23moved for the clinician.
39:25We see situations with heterogeneity
39:27where we have a clear 3 plus tumor where
39:31the patient gets you know trastuzumab
39:33and there's complete response and
39:35there's another tumor which was her
39:37to negative and was zero or you
39:40know one plus which didn't respond.
39:42So what will these patients benefit
39:45from a second round of DXD or
39:48when they have two distinct?
39:52Her two profiles to maybe I can use the
39:55microphone since there's 71 people online.
39:58Yeah, when I but I, I do want
40:00to clarify the question because,
40:01I mean we wouldn't use.
40:05We wouldn't have used a standard her
40:07two therapy if they were one plus.
40:13Why don't we choose?
40:18Do we use her two therapy, I mean. Yeah.
40:25But the patient?
40:29Patient had complete response
40:31to that through. So it was our CB0,
40:34but then the tumor which was
40:36one plus still extensive.
40:39Yeah so I mean it depends
40:41on the clinical situation.
40:42We know pretty clearly now that with
40:45before you know with the previous
40:47generation of her two therapies that
40:49you do not see any benefit with non
40:51her 2/3 plus or amplified cancer.
40:54So the her two lows do not respond
40:55to the previous generation any of the
40:57previous generation of her two therapies.
40:59So but with now with Tristan Madrox
41:01taken you know I think you could
41:04you know make a case that you know
41:06you might you would see potentially
41:08could see benefit both.
41:09And that clearly amplified in
41:11the her two lows.
41:12But prior to that we would look
41:14at a case like that on a case by
41:16case basis and say well let's use
41:18the her two therapy to get rid of
41:20that usually more aggressive her
41:22two her two positive cancer and
41:24then we'll worry about the her two
41:26negative or her two low cancer later.
41:28But it's it's you know again the
41:30the field is evolving now that we
41:32have these drugs that work across
41:34different levels of her too.
41:35I mean getting to in your earlier point
41:37that the two questions were brought
41:39up about her two heterogeneity and
41:40I think that's really interesting.
41:42Again with the first generation,
41:44her two therapies, it was very clear.
41:46We actually did a prospective,
41:47big prospective trial with with the other,
41:51the first antibody drug conjugate
41:53that doesn't have bystander effect
41:56and in that study a heterogeneous
41:58cancer responded much work much less.
42:01Effectively to a heterogeneous cancer
42:02than it did to a non heterogeneous
42:05cancer and and quantitatively that
42:07the to your specific question what
42:10mattered was the percent of her two
42:12negative cells not the intensity
42:15of her two on themselves.
42:18Again with a drug that has bystander
42:21effect as as I think David was alluding
42:23to that might be switched and maybe
42:25just if you just need to have a
42:27certain number of her two strongly
42:29positive cells to get the drug in.
42:31And then the and then the bystander effect
42:32will take care of the her two negative cells.
42:34We like to test out
42:35prospectively with we haven't.
42:38Had the funding yet to do that trial.
42:42Yeah. And I enjoy your talk, David.
42:46Is any do you have any information?
42:50The conjugate drug can get activated.
42:54In the extracellular and
42:56microenvironment of tumor cells.
42:58So I I would again defer to Ian,
43:01who's much more of an
43:02expert on this than I am.
43:03But it's my understanding that the drug,
43:05once it comes off,
43:06it has to be cleaved inside the cell.
43:09But once it comes off,
43:10it survives in the extracellular environment.
43:13And that's how the bystander effect works.
43:14That's how it can kill neighboring cells.
43:16Bystander effect doesn't really require.
43:20To go into the cells as long as it's a.
43:25Present in the microenvironment of the
43:28tumor cell in the enriched fashion
43:31you you will have some activities.
43:33Well the drug is an inhibitor
43:35of topoisomerase,
43:36so it has to get to the nucleus somehow.
43:38I guess to have its effect
43:40you can get activate.
43:42Outside
43:42of the cells. You don't have to take
43:45anybody to go inside the cells.
43:48Right, but that that that but
43:50the antibody doesn't necessarily
43:52take the the drug I guess can get
43:54into cells without the antibody,
43:55but the reason they conjugated to
43:57antibodies so you can increase
43:59the dose locally to the tumor.
44:02Michael environment.
44:03We have a more protease type.
44:07To break up the linkage between
44:11conjugate drug and the conjugate ohh,
44:13that'd be fine.
44:15That that that means you read that
44:18could partly explain why the the
44:23heterogeneity potential difficult
44:25involvement as well as you may
44:29have another interesting parameter
44:31to assess now today with the.
44:36Mass Effect. We could look into.
44:40Whether they're saying reached?
44:43Do you conjugate?
44:45Drugs.
44:45And in that case you would
44:46argue that it worked.
44:48It would work without trustors
44:49maybe even being present.
44:51You could you get the deconjugation
44:53even if there's no target present.
44:55That's that's why you get a negative.
44:57Right without right without her two
44:59present if if the if if it's a true her
45:0220 the drug could still work because
45:04it could get deconjugation and be effective.
45:07Everywhere in the body.
45:09Anywhere in the exercise room.
45:12Have unreached.
45:14That particular,
45:15yeah,
45:16I I think that you know,
45:17the the question is the toxicity
45:18then and that's actually the problem.
45:20I didn't go into that.
45:21But one of the problems with this
45:22drug is it has pulmonary toxicity
45:24and that patients get interstitial
45:26lung disease about 10% of the time.
45:28And that's another reason why
45:29you need a companion diagnostic.
45:31And one wonders if perhaps the
45:33interstitial lung disease is due to extra,
45:36you know,
45:37extracellular environment cleaving the
45:39drug even in the absence of her too,
45:42although we've also found
45:43that her two is present.
45:45Normal Airways at about the level it is in,
45:47low in low in about at about
45:491/4 UN quote one plus level,
45:51or between four and six
45:53animals per square millimeter.