Session II Transcripts
November 20th (PM) Session II: Pharmaceutics
Panel Discussion
Mike Hageman, Bill Charman, Brian Rohrs, Gail Strong
Question: I want to follow up with a people question that I asked this morning since
we now have the other half of the industry represented—Pfizer. On a serious note, my
perception has been that chemists seem to be more attuned to DMPK issues in terms of
optimizing those than necessarily physical chemical properties, in particular, solubility. I
find that curious. I don't know whether it's because they don't understand the kinds of
issues that you have brought up here this afternoon or because they feel that they can
pass the buck to all of you to fix the problem. I'm just wondering what your perception is.
I just feel that chemists are really attuned to what we talked about this morning, but less
attuned to this. So I'd like each of you to address what your experiences have been.
Hageman Response: There are a couple of reasons why that has played out that way.
One there's probably not too many cases where solubility clearly kills the compound that
they're trying to push forward. So I think DMPK information coming out often will have
greater ramifications than will solubility. A second is that trying to adjust solubility and
exposure is often opposite directions from the perspective of the medicinal chemist, that
is, they generally are thinking of things in terms of lipophilicity, and it becomes a very
tricky balancing act with respect to trying to enhance solubility yet maintain binding to a
receptor and/or transfer across a membrane. So I think there's a kind of natural
tendency to throw their hands in the air with regard to addressing that issue. The final
thing that has given us a lot of difficulty over the years is that there's always the $800
million drug exception to these issues that we continually argue with at the project team
meetings and we often find it very difficult to argue around.
Rohrs Response: I agree with what Mike said; there's a couple of other factors I can
think of off the top of my head. Typically with DMPK issues, there's a clear target
endpoint—if it doesn't meet this, that's it, forget it. With solubility issues and kind of that
exposure stuff, pharm sci is always going to view themselves as an enabling
organization in terms of these early lead optimization efforts. So there's always hope
that we can do something about it. And it doesn't fall in the chemists' lap to do
something about it, it falls in our lap. So I think it's perhaps not so much that they're
maybe not attuned to it, it's the fact that they can pass the buck, although this has bad
connotations, but there's the ability perhaps of a technology to save a compound, and I
think that's what they look to us for. Again, and I'm putting thoughts into their heads, I
don't think it's something they even think about because it's not something they have to
worry about specifically until they're told to. That's been my impression working on
those types of teams.
Borchardt Response: I've seen in the past year several examples where compounds
were going to get killed because the solubility was so low that you couldn't get
appropriate exposure for tox studies.
Strong Response: I would comment that I think that when we bring the issues of
solubility or formulation to our teams, one of the tools that we don't have readily
available is the ability to quantitate the costs of the approaches that we're taking or the
cost of dealing with a low solubility or a difficult to formulate compound. In terms of the
cost of the technology, the risk of the technology, the potential delay of getting that
compound to market, those all have an inherent cost that should be weighed against
screening for a more attractive molecule. But because it's more difficult to bring a "this
one will take 3 months longer," "this one will have a risk of failing 2 batches at phase 2
or phase 3," we can't quantitate those risks, so it's hard to bring that into perspective
with the Ki for this or the Ki for that. So we're missing some of those evaluation tools
that might bring this component of the whole drug development requirement into
perspective for your compound selection.
Bill Charman Response: One way I can answer is in terms of our ability to predict the
potential for lymph transport. Firstly, the proportion of compounds that go through the
lymphatic system is quite small as a percentage of all compounds. In terms of
prediction, we need to consider compounds that are highly lipophilic as estimated from
a high log P value and high triglyceride solubility. Also taking into account the potential
for the solubilization equilibria based on the lipophilic weak acid/weak base issue that I
spoke about this morning, we're reasonably confident that we can predict the likelihood
of potential transport. So for the compounds that fit those criteria, the potential for
transport is definitely quite significant. From the point of view of being able to
deconvolute humps and bumps in plasma profiles to make a suggestion as to whether
or not lymph transport may be a contributor to the plasma profile, I would caution you
not to do that at all. There are lots of reasons that one can get lumps and bumps in
plasma profiles after oral administration and lymph transport may be one of those
contributors. If you remember back to the data that I presented this morning, the
transport lymphatically was essentially complete within 3 to 4 hours. That's not going to
show up as a hump in a profile. And as I said, because of the different mechanisms
that contribute to the double peaking or humps and bumps, I think it's somewhat risky to
try to deconvolute those profiles to suggest whether or not lymph transport is
contributing to bioavailability. The third question was whether or not people are
intentionally utilizing lymph transport as an absorption pathway. The answer is yes,
some groups are.
Hageman Comment: I have one further comment with regard to a couple of examples
of what Ron had mentioned. One of the situations I've seen where solubility got driven
home pretty quickly was situations where we didn't have adequate solubility to run the
QTC studies. So, it wasn't because we needed solubility to have a more "drugable"
product. Again, it limited them doing something which, as Bill had just indicated, has
become a clear criterion that you must meet. So then they began to worry about the
solubility issue. The other one that Bill alluded to, which is a people issue, is this
education or communication. I still remember early on when I attended a project team
meeting and I just couldn't figure out what they were talking about. They introduced a
tert-butyl group onto the end of a functional group and they said it increased the
solubility. This went on and on and I just couldn't follow and, finally, after I asked
enough questions it came out that it increased the solubility in their fluid they were using
for crystallization, which was an organic fluid. So, it's just an issue of we're not even
talking the same language at all.
Question (for Bill Charman): As far as lymphatic absorption goes, how prevalent is
that do you think for compounds just across the board as far as their ability to take
advantage of nascent levels being made and being packaged and sent out into the
vasculature? And are there tell-tale signs in the PK when this occurs? Are there humps
later on in the absorption phase? Is there anyone out there—chemists utilizing
techniques to try to shuttle compounds by that pathway?
Bill Charman Response: One way I can answer is in terms of our ability to predict the
potential for lymph transport. The proportion of compounds that go through the
lymphatic system is quite small as a percentage of all compounds. In terms of
prediction, we need to consider compounds that are highly lipophilic as estimated from
a high log P value and high triglyceride solubility. Also taking into account the potential
for the solubilization equilibria based on the lipophilic weak acid/weak base issue that I
spoke about this morning, we're reasonably confident that we can predict the likelihood
of potential transport. So for the compounds that fit those criteria, the potential for
transport is definitely quite significant. From the point of view of being able to
deconvolute humps and bumps in plasma profiles to make a suggestion as to whether
or not lymph transport may be a contributor to the plasma profile, I would caution you
not to do that at all. There are lots of reasons that one can get lumps and bumps in
plasma profiles after oral administration and lymph transport may be one of those
contributors. If you remember back to the data that I presented this morning, the
transport lymphatically was essentially complete within 3 to 4 hours. That's not going to
show up as a hump in a profile. And as I said, because of the different mechanisms
that contribute to the double peaking or humps and bumps, I think it's somewhat risky to
try to deconvolute those profiles to suggest whether or not lymph transport is
contributing to bioavailability. The third question was whether or not people are
intentionally utilizing lymph transport as an absoprtion pathway. The answer is yes,
some groups are.
Question: Mike, in your talk you raised a major specter for those of us that work in
drug discovery, and that is maybe some of these bioassays are not being conducted
properly because of the solubility of compounds. So I have several questions regarding
that. One is: I can see how that would be a problem when we're screening—we might
be screening at, say, 10 micromolar and we might have compounds that don't have
quite that solubility, but when we get into the optimization phase we're hoping to get
down into the submicromolar range. So, why would that be a problem when we're
talking about, hopefully, nanomolar compounds. Secondly, what do we look for if we're
having that problem? For example, I can imagine you get flat IC50 curves or something
like that. But is there something else that people in the discovery teams can look at the
biological data and say, "Oops, there may be a solubility issue here"? Third, has
anyone looked at the idea of giving feedback to the biologists about how maybe they
should be doing experiments, or maybe some generic way to increase the solubility?
When I first encountered this where I work I was shocked that this hadn't been fully
understood by the biologists developing assays and high throughput screening people.
So is there any possibility that we could come up with a few generic additives that would
go into these bioassays that would tend to solubilize a greater majority of the
compounds so we can better biology?
Mike Hageman Response: In our experience, as you begin in the optimization realm,
where you start going down to the nanomolar type of IC50s, it probably is less of a
problem. Some of it depends on where you start out with your reference material, in
particular your DMSO stock solutions or whatever you might choose. Because, actually,
the biggest problem we have is that first dilution step. To some extent you're a bit
limited with volume issues to do much more than an accurate 1–100 dilution. I think
some of the newer equipment can push it more than that, but then you end up doing
subsequent dilutions anyway, and if you remember those curves in DMSO, it would be a
co-solvent system where it's an exponential relationship with regard to the concentration
of DMSO. So, one has to think about those curves a bit when you do those dilutions in
that it's good either not to dilute too far or dilute as far as you possibly can. Just doing
something in the middle is probably the worst scenario that you can go with. So, I think
you can work with people with regard to the dilution sequence and how you go about
approaching that. Another one that becomes a big problem with the nanomolar
binding—and we often had questions about that—is now you're starting to get into the
potential for a lot of things like nonspecific absorption to play significant roles and
provide false negatives. There probably are some additives and over time I think a lot
of the biologists might use small amounts of PEG or, in this case, casein. They use the
casein protein to prevent the nonspecific absorption totally not appreciative of the fact
that also there was some binding, although it was low, of the drug to the protein. So I
think there's probably some things that can be worked out there. One of the last
questions was about working with the biologists to come up with generic ways to try to
get around some of these issues. Our experience is that, generally, those things are
built by historical experience and most of the assays are built based on what has
worked for that group in the past. I think we did have a couple of pushes at Pharmacia
to try to get involved with the high throughput assay group and simply have somebody
at the table before they started these high throughput assays so we could at least see
what was going on and sometimes make some very simple suggestions. Because what
generally happened was that once they'd started the assay and had done their first
100,000 compounds, it was very unlikely that they were going to change it in any way
because it makes it difficult to compare the previous data. So usually we found
ourselves trying to add an additional assay to try to get the information, which usually
wasn't received very well.
Question: My question relates back to Ron Borchardt's question about the division
between the discovery and the processes that occurs later when the formulation really
tends to come into play. In particular, in cases of compounds with low solubility when
you end up dosing the animal—you might first try methocel and then try in water/Tween
to improve the absorption—the problem I see with this is that you want to work in the
discovery stage at concentrations that are near your efficacy range. You then go to
safety and you want to bump that up 30- to 100-fold. So the question I have is, are we
shooting ourselves in the foot by trying to accommodate the properties of the compound
early on, or do the use of some of the formulation approaches end up compensating for
that in the long run?
Mike Hageman Response: That's a question we continually ask ourselves: Are we
enabling poorly drugable candidates to move forward by introducing solubilization
techniques, formulation techniques, and so forth? I think if used in a vacuum and not
used with the appropriate education, that could be the final outcome. I think what's
more important in a way, at least in the interactions with the teams I've been involved
with, is trying to push it from the perspective of what is the question that you're trying to
answer. If that question is they can't get high enough levels to even help identify a
template in vivo—you know, we could go to extremes to try to get the exposure for that
so then they could start their analoging of a given template. I think there is that risk and
we discuss it often. The other thing is, obviously, if you begin to go up this path the cost
of development of these types of solubilized formulations is clearly higher. And as an
organization responsible for formulation development, that's something we'd prefer not
to engage in.
Gail Strong Response: One of the things it comes down to, as Mike was saying, is
what is the question you're trying to answer. Also, if the studies are done as part of a
team environment where you say, "OK, we're going to answer the question about firstpass metabolism by dosing solution of this compound vs. solution of that one," as a
member of the team you also have to determine is there a viable delivery-type
formulation that could be taken to the clinic and to the market. So just because you get
a positive with your solution or your solubilized formulation, that only takes you one step
down the path; it does not make a candidate drug. So, if you've got the team set up
such that you need to also demonstrate viability for a clinical or marketed formulation,
i.e., a suspension or some prototype tablet, then you don't get locked into these
formulations that answer the early question but put you in a path for a very high cost,
very high risk development.
Brian Rohrs, Pfizer, Response: Often you'll have compounds that could be helped—
Gail had a good case study with, say, salts and things like that—but at the time you
need this other information you don't have the salt forms available or you don't have
some of these other perhaps enabling technologies characterized well enough to be
available. So sometimes, in fact often, these solubilization technologies act as a bridge
in time between "we want to look at these properties" and "we think that down the road
perhaps we can solve some of these problems, but we don't have them solved right
now." So there's a temporal aspect to this that we have to consider.
Question: This question is for Gail. In the work that you presented on the use of
disalts; at what stage of drug development did you do these? Discovery, preclinical or
clinical?
Gail Strong Response: We did this work at the preclinical stage when we were trying
to prepare the candidate molecule to get into the first studies in humans. In that we
knew that the target product profile was a solid or a dosage form and, while we had a
solution formulation that worked very well for tox and did indeed go into the phase I
studies in volunteers, we started prior even to those clinical studies evaluating solid-type
formulations, of course, assuming success for the molecule, which we did not have, and
that's why we get to talk about it today. So the majority of this work started prior to
going into the clinic and was continued in parallel with the early clinical studies.
Question: Would you do this for every compound that shows up?
Gail Strong Response: No, but certainly in terms of compounds that have reached the
product candidate/development candidate/exploratory candidate status, where you
really have a commitment to get it into human studies then you need to make sure you
have an ability to take this molecule forward in the clinic beyond the first preliminary
formulation. And so we did some of that work prior to the clinic.
Question: I was only wondering if when you're facing discovery with many compounds
that come into your lab that you have to deal with, you don't have the luxury of is this
salt going to work or is that salt going to work, you just want a simple CMC or HPMC—
something that will make it go—and that's what you go for if it's agreed by your
company. I was just wondering if that's what you do every time, but you don't.
Gail Strong Response: I would comment that in terms of nomination criteria, and
again you hate to have hard and fast rules, you did not need to have a form of your drug
substance that would be acceptable for development. So, DMSO solution is perfectly
acceptable for screening and early pharmacology, but to demonstrate that this molecule
has properties that make it a viable or potentially viable candidate, we also needed to
demonstrate that we had a manufacturable API form and formulation at least options for
downstream clinical evaluation. And that was a component or criterion of acceptance.
Question: I was only wondering if when you're facing discovery with many compounds
that come into your lab that you have to deal with, you don't have the luxury of is this
salt going to work or is that salt going to work, you just want a simple CMC or HPMC—
something that will make it go—and that's what you go for if it's agreed by your
company. I was just wondering if that's what you do every time, but you don't.
Gail Strong Response: I would comment that in terms of nomination criteria, and
again you hate to have hard and fast rules, you did not need to have a form of your drug
substance that would be acceptable for development. So, DMSO solution is perfectly
acceptable for screening and early pharmacology, but to demonstrate that this molecule
has properties that make it a viable or potentially viable candidate, we also needed to
demonstrate that we had a manufacturable API form and formulation at least options for
downstream clinical evaluation. And that was a component or criterion of acceptance.
Mike Hageman Comment: I think that was also a good example of where it depends
on the therapeutic category and the goals of the corporation at that point in time. You
have to recognize this is a protease inhibitor coming at the time when everyone was
going after a protease inhibitor and resources were allocated in such a way. If it had
been the third or fourth generation of a cardiovascular agent for high blood pressure, I'm
not sure you would have seen that.
Question: Would you base the acceptance of going forward with a drug on solubility
alone? Because I can think of a number of drugs, methotrexate comes to mind, that
have horrible solubility but are very good drugs and have been used for ever. Secondly,
at what point would you consider IV injection over oral because of physical properties?
The third part of the question is for Gail. Do you have any idea why the amorphous part
of the HIV protease offered a better AUC than the crystalline form?
Gail Strong Response: I would say it's very unlikely that solubility would be
considered as the unique criterion for a go/no go decision on a molecule until you have
confirmed that solubility is a good predictor of the attributes of the molecule that you're
trying to develop. So if you have demonstrated that getting exposure or bioavailability
or efficacy is dependent upon solubility, once you have that correlation then you can put
a cut-off in for solubility. Without that information it's going to be difficult to make that
decision up-front. The second question was whether we would consider IV formulation
if we were not able to give oral exposure. I guess my comment would be that most
projects have a marketing component to them. They are identified in order to be
successful in this therapeutic area for this indication, we need oral or IV would be
acceptable, so I think much of that is driven by marketing. I'm going to throw in a kind of
side loop here. I vociferously oppose doing an IV formulation just to get absolute
bioavailability of a compound, really based on just cost and development. So I think the
route of administration is normally predetermined, really by marketing forces. The third
question was why do I believe that the amorphous material or the dispersed
formulations gave better exposure. The data, as I look at them, indicate that we're able
to get supersaturated solutions upon dissolution of those formulations, which give us a
higher driving force—get us up on that plateau—for a dose number that Brian talked
about to really maximize exposure.
Bill Charman Comments: The issue in terms of solubility, of course, is that it needs to
be put in the context of dose and potency. And that's probably the key reason that one
can't ascribe a particular absolute number to solubility alone because it needs to be
related back to potency and dose.
Question: Brian described a scenario where you can increase the absorption of the
high energy solid with high permeability/low solubility by reducing the particle size for
increasing the solubility of the formulation. Is there anything that you can do to disturb
the crystal packing so that you can use that crystal lattice energy or increase the
solvation? Are there any options?
Brian Rohrs Answer: As far as disturbing crystal packing, yes, amorphous materials
you create a higher energy solid because you're not packed in. It's pseudo-stable or
pseudo-unstable, depending on how far away you are. Obviously there are some things,
if they're big and greasy, it's very hard to get crystalline material period. Is there
anything you can do to disturb the crystal lattice? There's some work by Nair
Rodriguez-Hornedo in looking at co-crystallizations of various things that I think has
some potential—it's very early in the research—but it has some very interesting
concepts. By co-crystallizing different materials you actually increase the solubility of
one, the other or both of those materials. Other ways to disturb crystal packing are
through the salt forms, although that's maybe a little bit self-serving because you're also
ionizing the things, too, but that also disturbs the crystal packing.
Mike Hageman Comment: I think that crystal packing is one issue that people struggle
with a lot because one thing that they generally think about is trying to enhance
solvation by introducing the hydrogen bonding groups, which, in fact, can take you the
opposite direction with your crystal packing. So you do have to be careful about that.
One way that I've seen also is really taking a new approach—it would be a prodrug
approach—you can upset the crystal lattice and then obviously you can analog around
crystal lattice if you have some understanding of the crystal structure, which,
unfortunately, you don't usually have at that stage of the game.
Gail Strong Comment: This touches on a topic that's one of my favorites lately,
crystallization and salt selection. Certainly, the amorphous form, for example, where I
talked about the free acid, the crystalline suspension vs. an amorphous precipitate, yes,
the amorphous precipitate is meta-stable with respect to that crystalline material. At the
same time, when you make a salt of something, when that dissolves many times you
get a solution that is supersaturated with respect to the free base or the free acid. So
it's not truly meta-stable in that that solid form is not going to convert to the other one,
but when you go through a dissolution step you have the potential for precipitating a
lower solubility form. So it's important just to understand your boundary conditions and
understand what is your lowest solubility form, your most stable polymorph and then try
to decide how far away you want to be from that in terms of improved, say, solubility but
also potentially increased risk..
Brian Rohrs Comment: From a formulation approach you might also think about
something like a solid solution; it doesn't necessarily get to the "what can I do to this"
from the chemists' point of view, but from a formulation approach spray drying with a
polymer, trying to get a solid dispersion, is another way to disrupt the crystal lattice and
create that higher energy solid.
Gail Strong Comment: Yes, that was the approach we described for the 103017; it
was to disrupt the crystal lattice and then to stabilize it through the introduction of a
polymer. And that worked very, very well for us in that case. Again, though, there's
always the risk that that solid dispersion through exposure to humidity, to time, to
temperature could either dissolve and reprecipitate as a crystalline material or actually
rearrange in the solid state.
Question: This is more of a philosophical question following up on Jerome's earlier
query regarding providing solubility optimization support too early as an enabling
technology. Two years ago there was an FDA workshop on early results of the
implementation of a BCS system from a regulatory perspective. One of the more
interesting talks there was from the FDA, it was a pharmacologist, I think, who did a
retrospective analysis of the previous most recent 100 NDAs submitted to the FDA. He
classified all of the submissions in terms of their BCS rank and 50% of them were class
4s and fewer than 10% were class 1s, which would be high solubility, high permeability.
So they were predominantly low solubility, low permeability submissions. If one might
conclude from this that optimization campaigns were launched in all of those programs,
were beat to death and, at the end of the day, those were the only candidates that could
be submitted as potential drugs, are we by imposing unrealistic expectations and
objective in our discovery programs delaying the inevitable and adding additional time to
the drug discovery process rather than early on—what we're probably doing is
progressively lowering the bar in that eventually something is left standing that we can
develop as a drug. But if we accept that developable candidates may be available
earlier on and don't meet all of our early expectations of ideal, well-behaved drugs and
identify those as development opportunities earlier on, would we save ourselves a lot of
time?
Bill Charman Comment: I guess the comment I'd immediately make is that the issue
of the BCS classification is the solvent is either water or 0.1 N HCl. And the relevance
of that solvent with regard to the intestinal milieu, and of course the absorption of poorly
water soluble drugs, needs to be considered. So I think that that issue in terms of
classification, especially as Class 1, is an unreasonable one with regard to compound
selection at this early point. Clearly, for compounds to have got through the approval
process they would have sufficient reproducibility in plasma profiles, sufficient
reproducibility in terms of clinical response. So I don't think that's necessarily wrong; I
think it's probably the choice of solvent for the BCS because it's really looking at
something different with regard to compound selection at an early point.
Question: As a result of what we are asking to come out of discovery programs—and
this is something we discuss often, too—at what point do you reach a state of paralysis
with regard to trying to identify candidates that you can move forward? And, ultimately,
as we get all of this information, we may end up with some really good candidates, but
very, very, very few of them. Not that I have any answers, but I've sometimes seen
teams where we simply have way too much data and even to try to sort through it we
have some problems. I think there's a need out there for people to be working on how
do we handle this much information, especially as these high throughput systems have
come on.
Question: Gail, for the example drug that you presented, what was the formulation for
the acute and chronic toxicology studies?
Gail Strong Answer: We used 100 mg/mL dissolved in pH 10 NaOH solution.
Question: That was acceptable to the toxicologists?
Gail Strong Answer: It worked, yes, that is what we used. We inched our way up
there, I will say, and that was not met with great joy when initially proposed. But we
were able to demonstrate again through data and through experimental studies that it
was well-tolerated and gave excellent exposure and that is what we took into the clinic,
so that went into humans as well.
Question: It's been my experience that they never wanted to try anything that had pH
greater than 8.
Gail Strong Answer: I will agree with you that people are often very reticent to go
places they haven't been before and it's our job to discuss it and to present the data and,
I'm going to say, be persuasive. And you can be; it does work on occasion.