About this site
Here we'll review recent developments in drug discovery and medicine and the IP issues and financial implications they have, along with general thoughts about research. Also likely to make an appearance: occasional digressions into useful topics like which lab reagents smell the worst.
About this author
Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases.
To contact Derek email him directly.
Palladium, My Pal
I set up one of the largest reactions I've run in a while today. It's not large at all by ordinary measures - ten grams, not even half an ounce. And it isn't large by organic chemistry standards, either (the folks across the hall from me run hundreds of grams all the time.) But by the standards of my recent year or two of lab work, it's large. I'm not in the lab all the time, and when I am, I'm generally trying out small exploratory stuff. But this is an intermediate that I need to try out a lot of new things, so I thought I'd get it over with.
I'm not giving anything away if I tell you that it's a palladium coupling reaction. If I had to name the biggest difference between chemistry in my grad school days (mid-1980s) and now, it would be palladium chemistry. I should add to that the whole constellation of transition metal reactions (nickel, iron, cobalt, indium - all kinds of things get into the act.) New ones are being discovered all the time, along with new variations on reactions that had looked well-explored.
These things are good at forming carbon-carbon bonds, which are the hard currency of organic synthesis. C-C bonds are the backbone of all our molecules, and without some help (from other functional groups) they're basically unbreakable under controlled conditions. Now, uncontrolled conditions will work fine, though, like a big fire in your hood. But that's not a very useful reaction (unless you like plenty of paperwork, and maybe were looking to be named the next chairman of the Safety Committee.)
The reactions that the transition metals can do would have looked like witchcraft to most chemists a few decades back. These transformations have been gradually creeping into use over the last 30 years or so. The first people to explore this stuff had no idea what a large continent they'd landed on, and the early uses of the chemistry weren't always the ones that caught other people's interest. Over time, though, it became clear that this was the way to go.
One major rap against them, though, is that there are so many different ways to set them up. What metal to use, what form it's to be used in, what solvent (or mixture of solvents,) temperature, base, additives - all of these can vary all over the place. I think that almost any metal coupling reaction can be optimized to a high yield, if you're just willing to spend too much time doing it. Mostly, people like me (or one of my labmates, who worked this one out) just vary conditions until it works well enough, and then stop. We declare victory and get out, like they told Lyndon Johnson to do in Vietnam.
The other complaint is that it can be hard to get the last traces of metal out of the products. For medicinal chemistry, that can be a particular concern. Nickel's actively toxic, and palladium, while not as bad, won't exactly make your coat glossier. These reactions get used on large scale to make drugs (there's no other way to do some of these transformations,) but only with a lot of spadework to find ways to run them in the cleanest (and most easily cleaned) manner possible.
As for my reaction, it was a becoming shade of reddish-brown when I left the lab. There aren't many reddish-brown wonder drugs out there, so obviously that stuff will have to come out before I go on too much further. There's about fifteen grams of white crystals hiding in there - or at least, there had better be.
Meanwhile, over at Schering-Plough, Fred Hassan has taken over, as expected. He seems to have done a good job at Pharmacia (these things are harder to read than press clippings would have you think, though.) He'll have quite a job at S-P, but he has the advantage of having vast heaps of bad news land on the company before he got there. Things would have to start looking better for almost anyone; a quarter without a nasty surprise of some sort would be a real step up.
I've seen a report that Hassan has said that he wants to end Schering's dependence on single blockbusters. Said blockbuster would be Claritin, I assume. And most of the cold-turkey has already been done for him by the patent office and the market.
But I know what he means. Companies often talk about having a more balanced research portfolio, and about not leaning too much on one or two big sellers. They especially talk like this right after they've just lost one. It's an admirable goal, but no one will complain if a multi-billion-dollar compound shows up, no matter what the official policy is. And once one does, they'll lean on it for all it's worth, and who could blame them?
I don't think it's possible for a company not to become dependent on a blockbuster, and thus not to feel its loss when it (inevitably) goes away. It's human nature. It's also human nature for a CEO to avoid saying things like: "Hey, we'll take what we can get. Little ones, big ones, whatever you got. If we find a whopper, we'll sell it until we turn blue. If we can't get one of those, we'll do the same thing to whatever looks second best. Third. Eighth. Whatever." That's how it works, though.
More on Pfizer
Here's a take from the Wall St. Journal on Pfizer's acquisition of Pharmacia. Notable is the announcement of job cuts, which had been expected (and then some.) They're pretty extensive, but not as bad as some of the more frantic rumors of the last few months. This time, at least, it's not just the acquired company's employees who are being trimmed. A key quote:
Some observers wondered how Pfizer, already grappling to improve productivity in its enormous research operation, could manage after the addition of Pharmacia's sizable scientific enterprise. Part of the answer appears to be that the number of laboratories will shrink.
That'll help, up to a point. Can't shrink them too much, though, or too often, or you end up cooking and eating your seed corn. The rest of the answer remains to be determined, as far as I can tell. As does the relative size and importance of the job-cutting component - how close does that get them to making the whole thing work?
You Are What You Order
Longtime readers of my previous blog may remember that I was running a series of experiments last year that I had great hopes for. I spoke about these several times, but the problem was that I couldn't go into details about just what those reactions were. It's neat chemistry, and it could have some very interesting uses, so I have to clam up right there. (Working in industry will do that, but I wouldn't have been able to talk about these if I were in academia, either, for fear of someone scooping me.)
Well, I'm at it again. For the past few months, I haven't had a chance to work on that idea, but now opportunity is knocking once more. This is the same general idea as what I was doing before, but in a modified form (closer to what's known already - maybe that'll give it a better chance of working.) I hope to head out to the frontier with some of the reactions I was trying before, too, but what I have going now is wild enough by the usual standards.
And I'm still not going to be able to talk about it in any useful detail, which I realize is pretty irritating. So I'm not going to be updating this topic too frequently, just when crucial experiments are coming up (so everyone can do the agony-of-defeat thing with me, which is basically what happened last time.) But I wanted to let people know that I'm still doing this sort of thing.
It occurred to me the other day that just seeing what sorts of reagents I'm ordering would be enough to tip someone off, if they were well-versed in the literature. They wouldn't know what application I'm working on, but they'd know roughly how I'm trying to do what I'm doing. That's why such information is confidential: the various chemical suppliers are sitting on a lot of valuable information. They sit on it very well, too, since they know what the consequences for their business would be if things started leaking out.
In a previous job of mine at another company, we ordered so much of a particular chiral amine (from one of its only suppliers) that they got curious about us. When our first patent application eventually published, within a very short time we were in one of their ads on the back of the Journal of Medicinal Chemistry, promoting the useful properties of their chiral amines. "Has been recently used in the preparation of such-and-such antagonists. . .", footnote to our patent. They appear to have been keeping an eye out, wondering what we were doing with all that compound. I told people that I'd made the cover of J. Med. Chem., which always got a puzzled look in response (the cover's actually been the same molecular artwork every issue for a long time now.) I would explain that it was the back cover, which of course didn't help much.
Pfizer's Shell Game
I alluded the other day to Pfizer's recent growth by acquisition. There are some things about it that disturb me, so I thought I'd get them off my chest.
The first thing to realize is that Pfizer hasn't been buying companies as much as they've been buying drugs. Their takeover of Warner-Lambert was completely motivated by Lipitor (a strong motivating factor, to be sure, given its massive sales.) W-L had been a takeover target for years, and any such move would have surely complicated the deal that Pfizer had to co-market the drug.
Rather than take any chances, they decided to buy Warner-Lambert themselves, and get all of the Lipitor revenue rather than half of it. It's true that they also got some cancer expertise and some pipeline projects, both of which I'm sure were welcome. But make no mistake, this was the Deal To Buy Lipitor, and wouldn't have happened without it.
Pfizer's next move was to buy Pharmacia (which was several years past being Pharmacia and Upjohn.) This, in its turn, was the Deal to Buy Celebrex. That COX-2 inhibitor was originally a compound from Searle/Monsanto, who Pharmacia swallowed up (later tossing out the rest of Monsanto) in an earlier Celebrex-driven deal back in 2000. It was a major revenue source for Pharmacia, which they were using to try to get the rest of their portfolio going.
But now it's more coal in the furnace of Pfizer. And now we've come to the heart of my problem with their growth strategy. As the company gets bigger, these blockbuster acquisitions have comparatively less and less effect. It's a mathematical problem: how are they going to keep their earnings growth? How many more huge-selling drugs are available to be bought (and at what price?) and how much good are they going to do?
It's a Red Queen's Race, trying to grow like this in the drug industry. The money that's going to be made off a drug is largely made during its period of patent protection, and patents are wasting assets. (These days, they waste pretty quickly, given the regulatory time on the front end, and the sudden flight to generics on the back.) So Pfizer's big sellers will burn brightly but briefly, which isn't something the company doesn't know. So what are they going to do about it?
I'm sure that one idea is to turn these drugs into franchises, like AstraZeneca has been doing with Prilosec/Nexium. But, frankly, I think that the people who pay for the drugs are catching on to that strategem (witness the performance of Clarinex, son of Claritin, for Schering-Plough.) And it's no sure thing that you can even get your follow-up drug to work, or get approved (ask Lilly about the attempts to follow up on Prozac, but be prepared for some foul language.)
Of course, Pfizer could just up and discover some new drugs on its own (there's a thought!) After all, they're now such a much larger company and all. But the problem is that the one thing they need, R&D productivity, does not scale. If anything, it may scale negatively. Large companies do not, adjusted for size, discover or bring to market more drugs than small ones. Not that I can see.
So we may see Pfizer out shopping some more in a few years (or sooner than that if anything untoward happens to any of their big sellers. You never know when one of your major compounds is going to take a hit - ask Bayer about Baycol.) The price of these blockbusters is going to keep going up. Their effect on Pfizer's bottom line will probably keep going down. And eventually, the music is going to have to stop. If anyone, at Pfizer or anywhere else, thinks differently, I'd like to hear their reasoning.
The Way It's Done
Richard has an interesting Living Code reply to my Tuesday post below. Under the heading "Bad Model," he says:
There are certain therapeutic areas whose heads are directly responsible for how many molecules enter pre-development each year (roughly moving out of basic research into animal models in preparation for human trials). That is, these heads get to keep their jobs if they keep moving molecules into pre-D. If they fail for too many years in a row, out they go.
True, true. That's certainly the way it is at many large companies. And the results can be as he depicts them - an aversion to risk, for one thing. Risk aversion is a tough position to try to live with in the drug discovery business, because the whole endeavor is founded on crazy risk-taking. That can lead to an unrealistic expectation of how much risk can actually be taken out of the process, which leads either to hubris or to a feeling of inadequacy (depending on the self-awareness of the person involved.) Most methods to minimize risk don't end up working very well. You can end up like a hybrid of Sisyphus and Canute: condemned to roll your rock uphill, over and over, but only by telling it to roll by itself. Not too effective, on any level.
The research goals Richard talks about necessarily lead to very strict project timelines, which on some level are a necessary evil. You want people to move things along; a company of sloths that never deliver won't make it far. (Programmers and engineers know just what I'm talking about. You can tweak and polish forever if no one stops you.)
But just because you can screw up in one direction doesn't mean you can't screw up in another (a general principle, that one.) Telling people to deliver six months early is not an effective motivational tool for the mice, the cells, or the proteins. They can't be made to feel wahtever sense of urgency the people around them might feel; they'll do what they want to, same as always.
As an artificial deadline approaches, experiments take on a different tone. Anything that looks like it'll take too long to yield data doesn't get done, for one thing. Under pressure, there's a natural tendency to draw conclusions too hastily, and to overlook things that might turn out to be complications.
And finally, timelines, like research goals and quotas, get turned into political objects. People pad them or lowball them, depending on their position. It eventually gets to the point where the "expected completion date" of a project is the one time that it's guaranteed not to be completed.
So, to return to what started my conversation with Richard, I think we'd agree that there are some clear signs to look for in Novartis's new research site - that is, if they're serious about having a new attitude. If they have research goals to meet - this many new projects, that many clinical candidates per year, make your goals or we'll find someone who will - then it's a sign that no matter what they say, they're doing it just like the other large companies.
Degrees of Degrees
An oddity about medicinal chemistry is that there aren't many graduate schools granting degrees in it. And another oddity is that the graduates of those programs don't get hired with any greater frequency than the people with plain organic chemistry degrees. You'd think that specializing in the field you're being hired in would give you an advantage, but as far as I've seen, it doesn't.
One reason this happens is that such graduates risk being despecialized - which also seems a bit weird, since you'd figure that med-chem is a subset inside organic chemistry itself. But keep in mind that in a drug company, chemistry and biology functions are pretty thoroughly separated. The biologists never make the new drug candidates, and the chemists never test them: not on proteins, cells, and certainly not on animals. (That last category is a specialist's job within a specialist's job.)
In an academic medicinal chemistry program, though, the lines can't help but get blurred a bit. After all, they want to give everyone the whole experience. Med-chem students sometimes end up doing both the chemistry and biology. It's a nice way to get a personal feeling for the science, but it comes at what's at least a perceived price: people end up as pretty good chemists, for biologists - or pretty good biologists, for chemists.
But that's not enough. There's no call in most drug companies for an organic chemist who can step in and run binding assays. It just doesn't come up. Companies would rather have people who spent more of their time on organic synthesis, and can crank out the molecules (and the ideas for molecules.) The biology is something that you can pick up after you're hired. People do so during their first few years in the field, if they're any good.
Conversely, if you run into someone who's been doing med-chem for a while and still doesn't have a clear grasp of basic pharmacology, then you should be suspicious. They're not doing their job, although I have seen people from some highly compartmentalized companies that end up like this. From the perspective of most other places in the industry, though, an experienced med-chemist without a feel for biology is someone who's been crippled in their development.
More On Novartis in Cambridge
My fellow Corantean Richard Gayle posted his take on the Novartis Boston Globe article I discussed the other day. He's picked up on another point of the article, that Novartis appears to be trying to change their corporate culture at the new site.
The description of the current Novartis culture is pretty interesting, if it's true. The original Globe article says that the Cambridge facility will "will look very different from the Basel operation, where labs intentionally were designed for secrecy, both within the company and without." Now, secrecy from the outside world should go without saying. Anyone who works in the industry has had those notes from management saying "Clean off all those structure drawings from the glass of your hoods; wipe down the whiteboards. We have visitors coming."
But secrecy within the company? That's a terrible mistake. From what I'm told, though, that's not really how they've been running things over there. You can't run research on a need-to-know basis, not for long. I mean, no one wants to get up and talk about just how terrible their project is doing, for example, so there are project managers who try to put off the day of reckoning. But in the end, the cards have to be on the table. At the companies I've worked for, anyone who wants to find out what's going on in another project can do so. (Of course, people are usually too busy with what's in front of them, but that's another topic.)
But on other levels, the Novartis site is certainly being billed as a hybrid of small-biotech style and big-pharma resourcing. It'll be interesting to see if they can pull that off. Others have tried. It's true that there are a lot of things about Big Pharma's culture that could use a good hosing down, if not a good sandblasting. Read Richard's article for the details, and the take he has on things from his time at Immunex. Having never worked for any small research firm, biotech or otherwise, I can't bring the same perspective to it that he does.
Here's what worries me: If the reasons given for the change (the overhyped version of the Gleevec story as presented in the article) are the real ones, then I worry that the entire idea might prove to be just as specious. I really don't think Gleevec was that revolutionary. Not that Novartis didn't do a good job developing it so quickly, but I see it more as a difference of a degree than a difference of kind. The ideas that Novartis are pursuing are worth more than the article makes them out to be, and for different reasons.
Part of my scepticism comes from personal experience of wave (after wave) of dang it-we're-going-to-change-the-way-this-company-does-business schemes. So far, none of these have quite worked out. The culture of any large institution is a very difficult thing to alter, a problem that's been recognized for a very long time without any good solutions presenting themselves. The phrase "No More Business As Usual" has a way of growing a comma after the word "no."
And the description of Novartis's new site ("open lab spaces and canteen areas, features meant to encourage the 400 scientists who will eventually work in the building to socialize and share ideas") sounds suspiciously like a new building I moved into some years ago. It was billed as a "high-interaction" facility, and although nicer than the old pile, high-interaction it was not. There were plenty of people that I hardly saw again after the move. In the old buildings, the mazy architecture that had built up meant that you had to go past everyone's labs in the course of the day. The new building spread everyone out between several floors, with no reason to travel between them.
But bad experiences aside, I'm still deeply sympathetic of the goals that Novartis is talking about, and the ones that Richard outlines. For example, getting people out of meetings and back into the lab - or at least at their desks, for starters - is a cause that I'll always rally to. I'm known at my current workplace (and was known at previous ones) for missing as many meetings as I possibly can.
The fight against top-down decision-making will always win my heart, too. I've long thought, perhaps too idealistically, that if you give people more power than they thought they could have, then the best ones would find a way to use it well. Of course, others will find a way to bungle things, but what of it? You've then identified those folks (a public service all by itself) and can, in theory, remove them from positions where they can do harm. Think of the way that good military commanders rise to the top in wartime, under powerful selection pressure. How are you ever going to find out what your people are capable of if you don't let them make real decisions?
I'll watch the Novartis experience with interest. A good example there could have an effect on plenty of other companies. . .
The ABCs of Drug Transport
A recent New England Journal of Medicine article highlights one of the ways that cells can insure that a drug doesn't work. In this case, it's medications for epilepsy. For some (but definitely not all) forms of that condition, the existing drugs can do a pretty good job of controlling seizures. But even among patients that could benefit, up to a third of them don't respond.
Researchers in London have found a bit of the answer. Studying the genetic makeup of nonresponders versus responders, they found that a particular form of the gene for the transport protein ABCB1 was expressed at a higher rate in the first group. This genetic variant, a one-base switch in the usual DNA sequence, causes the protein to be more heavily expressed in cells.
The association with this marker makes sense, because that protein is part of a family of efflux pumps. These proteins are found at the cell surface, spanning the membrane. Their working day is spent yanking various molecules from the inside and tossing them to the outside. (The similarities to a bouncer at a club have been often remarked on.)
What's odd about these proteins is that they have simultaneously very broad and very precise selectivity. A given efflux pump will respond to a wide range of structural types - wide enough that if you show a list of affected drugs to a chemist and ask what they have in common, you're most likely going to get a bewildered shrug. But at the same time, they don't just pump out everything. They're just very good at noticing compounds that aren't regular features of the neighborhood.
These things are responsible for a host of otherwise-inexplicable failures in the drug discovery world. Bacteria and fungi express them - and swap DNA plasmids for them like trading cards, a key cause of antibiotic resistance. The proteinss are found in the cells in the gut wall, and can cause compounds that you'd think would be taken up from an oral dose to be thrown back. They're in the epithelial cells of the brain's vasculature, which are hard enough to penetrate as it is, pumping attempted CNS drug candidates away from the brain. Most famously, they get upregulated and heavily expressed by the most intractable cancer cell lines, causing most attempts at chemotherapy to be tossed right back out. (For a review, see this article.)
But the epilepsy story isn't as clear as some of those examples. This variant of the ABC transporter is a contributing factor, but not the complete answer by any means. The variant still only showed up in 28% of the drug-resistant group, so the others are failing to respond for some other reason (and you can't rule out another transporter protein.)
More oddly, 16% of the patients who actually respond to the drugs turn out to have the same variant. The difference between the two groups is statistically significant, but that figure does leave folks searching for explanations. The authors themselves note that there's a lot of variation in that particular gene, and that they may have found something that's not the cause itself, but just linked to it. I think they're right - they've stuck the truth a glancing blow.
Novartis and the News Flash
Novartis has been reaping publicity for its move to new facilites in Cambridge, MA. Part of the reason is that this expansion is so obviously a break with company tradition (not Basel, not New Jersey.) And part of the reason is that there isn't any major expansion going on in the drug industry right now (quite the opposite at some places.)
But some of it is hype. In that category I have to reluctantly put a recent Boston Globe piece, brought to my attention by a reader. It starts off with a quick highlight reel of the company's successful kinase inhibitor, Gleevec. Daniel Vasella, Novartis's CEO, reviews the impressive phase II data for the drug and is quoted as saying "A moment like that comes maybe once in a lifetime." There's actually no exaggeration there. I've been at this for over 13 years myself, and I've never worked on a compound that's had a successful Phase II trial. To have one that not only works, but works to such a startling degree is indeed a rare event.
But from there on, the Globe piece starts going off the rails. Vasella is quoted again as saying that "Gleevec was the first designer drug for cancer." Which is true. . .but only up to a point. Gleevec was actually first targeted cancer drug that ever worked, the first to make it through trials. There had been many attempts, by many other companies, and many of them had failed. Some of the others were still working their way through the clinic as Gleevec made it through.
So to go on, as the article does, about how "Gleevec. . .was a catalyst for revolution at Novartis. . (it) convinced Vasella that Novartis should change its approach to science itself." Wow! And this new approach is? Well, first, let's hear about how things were done up until Gleevec:
"Traditionally, drug researchers mixed chemical compounds and tested them in animals with the hope they would treat a disease. Often scientists didn't know how the disease, or the drug, actually worked. The failure rate is high."
If by "traditionally," we mean "as of 1970," this is fairly accurate. As a characterization of the recent state of the drug industry, it's close to libelous. The picture you get from this is of researchers just randomly banging out compounds, mechanically stuffing them into white mice, and sitting back, pen in hand, to see what happens. Even a blind pig finds an acorn every once in a while, so in the fullness of time, the lab-coated bumblers eventually trip over something that actually works. Stepping through the thigh-high piles of deceased mice, they announce the discovery of another drug - while in the background, the lab assistants get back to the daily grind of decant-and-inject. . .the search goes on. . .
What a load! The reality is that everyone in the industry has been designing drugs around molecular targets for a very long time now. You have to go back well before I got into it to find people from the previous era. That's not to say that there aren't drugs out there that came along without their mechanisms being known. Metformin (Glucophage) is a good example of one (its mechanism is just now being figured out - we think,) and Schering-Plough's cholesterol absorption inhibitor (Zetia) is another one. That latter one, though was discovered by accident while Schering's researchers were targeting another specific enzyme in the cholesterol pathway; it wasn't something designed from the beginning to take advantage of random chance.
Those examples aside, I've never personally worked on a research project that didn't have a specific rationale. They generally go something like this: "We believe Enzyme X is important in the biological pathways of Disease Y, for the following list of reasons. . .therefore, inhibiting Enzyme X should be a viable target for the disease. It's worth high-throughput screening, and it's worth putting medicinal chemistry on if we find a starting point." If you want to propose a research plan that says "We're going to put compounds into diseased rodents until we find one that makes them better," I advise you to work on your resume first. You can try pointing out that the Boston Globe says that this is how the rest of the industry works, but that won't buy you much time while they hustle you out the door.
Don't get me wrong - I have nothing against Novartis. They have some very good people there (I know several of them personally.) And Daniel Vasella, whom I certainly don't know personally, is definitely no fool. I'm fond of a quote of his from a few years ago: "If you don't want to spend the big money and take the big risks, you shouldn't be in the pharmaceutical industry." He's as right as can be. And I wish Novartis luck with the Cambridge site, because anything that helps to lift the industry out of its hiring slump is fine by me.
But to pretend, or have a newspaper article pretend for you, that target-based drug disocovery is some sort of new paradigm is just disingenuous. We're all doing that; we've all been doing that for a long time now. And guess what? Just like the article says, the failure rate is high. If Novartis - or any other company - finds a way to measurably decrease that rate, they'll rule the world.
Elan Rides Again
I wrote several times about Elan's attempt at an Alzheimer's vaccine back on my Lagniappe site, the precursor to this blog. The admittedly long-shot idea was that the beta-amyloid protein plaques that seem to characterize the disease could be cleared out by an immune response. There hadn't been many serious attempts at that, since the brain is supposed to be something of an immune-privileged organ, but in rodents the technique seemed to work spectacularly well. Now there's more news about the attempt in humans, but whether it's good news or bad news is, well, a good question.
The April issue of Nature Medicine has a report on a detailed study of one of the casualties of the vaccine trial. That's an accurate description, unfortunately. Several of the treated patients became ill with an inflammation of their brain tissue, strongly presumed to be linked to the vaccination program itself. The trial was halted in early 2002, and some of those affected made at least partial recoveries. But some didn't, and this particular patient declined severely and died within a few months of the last vaccine injection.
A detailed examination of their brain tissue has provided plenty of information, but not many definite answers. (There's no way that it could provide these, in a general scientific sense, because this is an N of one. We have no similarly affected brain to study, and no control from someone who was in the study and unaffected.)
Those cautions noted, here are some of the findings. For one thing, substantial portions of the cortex appeared to have been cleared of amyloid plaques. But not all: the distribution is reported as "patchy," and there doesn't seem to be a good explanation of the pattern that's seen (for example, the frontal lobes appeared to be untouched, and as full of plaques as a standard Alzheimer's brain would be.) One problem is that we don't have a clue of what the plaque distribution was like in this patient before treatment - how patchy was it to start with?
And another hallmark of the disease remained, the neurofibrillary tangles. That's no particular surprise, since the vaccination wasn't targeted to remove these. And it's very much an open question which pathology does more damage to brain function. (Whether anyone's trying a vaccine approach to the tau protein found in the tangles, I don't know. . .)
There were also many signs of inflammation, which was expected. Some of the microglia cells that were involved in the immune response seem to have fragments of amyloid protein in them; perhaps they were caught in the act of clearing it. What isn't known is how all this differs, in quality and quantity, from the patients that weren't so badly affected. And Elan hasn't released any cognitive function test data from the trial, so we don't know yet if any of this did any of the subjects any functional good. (Rodents show improved memory function after the vaccine treatment, but that's a long leap - not that there's any good alternative.)
Immune responses can be massive and destructive things; the only time you'll see them seriously advanced as therapy on a person's own tissue are for things like cancer and Alzheimer's. As I've said before, immunity is one of those things like calling up demons in medieval black magic: never summon up anything that you don't know how to send back down. For example, one worry in this approach was that the immune response might start targeting the larger protein that beta-amyloid is derived from. That would be very bad news, since this precursor is found in virtually every tissue in the body. Another potential problem is that Alzheimer's patients tend to collect large amount of beta-amyloid in the walls of cerebral blood vessels. That's certainly not a good thing, but it could be even worse to suddenly clear it out. What if the amyloid is actually holding some of these vessels together?
So what's the path forward? One approach is to come in with a different peptide for vaccination. There's some evidence that one section of the full-length beta-amyloid elicits an antibody response, while another brings on the T-cells. If you could get the former and not the latter, the inflammation problem presumably wouldn't occur. Alternatively, you could just immunize with antibodies directly. Those don't cross into the brain well, but there's some evidence that if you clear the beta-amyloid from the rest of the body, that the protein in the plaques starts to diffuse out - sort of opening the drain and letting the chemical equilibrium do the job for you.
There could be trials of these approaches later in the year. If one of these trials goes off well, it's either going to be a huge advance in Alzheimer's therapy, or it's going to put a huge hole in the theory that beta-amyloid is a cause of the disease. Stay tuned.
Pfizer Takes Over
The last regulatory hurdle seems to have cleared for Pfizer's acquisition of Pharmacia. I'm a bit sad about that, for reasons that I'll go into soon. (I'm not a big merger fan in general.)
It also clears the way for the expected move of Pharmacia's former CEO, Fred Hassan, to join Schering-Plough. They've been looking for one for some time, and it's been widely rumored and reported that they've come to an agreement with Hassan. If this doesn't come out as a news item within the next week or so, you'd have to wonder if something's wrong.
One of the things that'll drive a medicinal chemist crazy - OK, some of us have a shorter commute than others - is polymorphism. The multiple shapes that the term refers to are the cystals of our drug substances, oddly enough. You wouldn't think that such a thing would matter, but it can be a complete deal-breaker in clinical development.
Most small-molecule compounds will crystallize under the right conditions. (The exceptions tend to be things that have tremendous flexibility, with lots of long, floppy chains - think vegetable shortening. But we don't turn out many molecules like that in the drug industry, because we're usually interesting in things with more defined shapes that can fit dependably into their molecular targets.) Crystalline compounds are a good thing, because they give you reliable forms of your drugs, with defined melting points and other physical properties.
Until they don't. The problem is that most substances can adopt more than one crystalline form, which all comes down to the order that the individual molecules pack in. Some of these arrangements are notably tighter or looser, which leads to higher or lower melting points in the bulk solid. The same effects often lead to different rates of dissolution when the compound is taken up in solvent - or, more to the point, when it's swallowed and sits in the stomach.
And there's the problem. A lot of work gets done trying to optimize the formulation of a drug, trying to make sure that as much of it as possible gets out of the gut and into circulation. It's a black art, for the most part, with all sorts of tricks and potions. Grinding the compounds to different particle sizes helps (and there are several ways to do that,) and so does adding surfactants and dispersing agents (and there are dozens.) But all that tweaking goes for naught if you suddenly start producing a different polymorph.
What was once tiny rhomboid-shaped crystals can decide to turn into needles with a melting point that's off by thirty degrees. To add extra enjoyment, this sort of thing usually happens as the compound is being seriously scaled up for larger studies. That's when the chemistry often changes, as the methods that the med-chem labs used gets adapted for more demanding conditions. Solvents and temperatures get altered to things are are acceptable on large scale, at the very least, and fairly often the entire synthetic route has to be reworked. Any of these changes can cause the compound to find a new way to come out of solution.
I know of more than one promising compound that has been delayed in its development because of a sudden change. When that happens, getting back to the original material can be a challenge. This is the sort of thing that made Rupert Sheldrake come up with his idea of "morphic resonance," some sort of weird shape-determining field that makes things assume the forms that they do. I think Sheldrake's full of fertilizer, but I can see where some poor process chemists might be willing to go along with the idea. Once a more stable polymorph is loose in a lab, tiny amounts of it can contaminate things and nucleate its formation under conditions which didn't produce it before. (The similarities to Kurt Vonnegut's "Ice-9" from the novel Cat's Cradle have been noted.)
But there are plenty of counterexamples, compounds that can be consistently produced in one form or another depending on how they're treated. What you don't want is a process that sometimes makes one form and sometimes makes another. That's irritating on a small scale, but disastrous for real production.
It's a global-minimum problem, and those are famously painful. Somewhere on the energy surface there's a sinkhole representing the most stable crystalline form for a given compound, but you've no way of knowing if the hole you're in is the deepest one. There are ways to help determine if different polymorphs exists, and various voodoo tricks to try to equilibrate them, but the uncertainly can persist until several large batchs of compound have been prepared.
Sequencing and the March of Time
Congratulations are in order to the team that sequenced the SARS candidate virus, of course. Talk of this producing a drug is premature, as I've been saying, but it's an excellent start for the vaccine people. The whole effort, though, shows what sequencing has become: it's Something We Can Do. Most microbial sequencing efforts don't even come close to making headlines these days. It's becoming a routine tool.
Don't think that I'm complaining. No one misses the days when we didn't have this information. You can be sure that the molecular biologists don't long for the era of practically doing the work by hand. I saw people chopping cotton with hoes when I was a young child; mechanization doesn't bother me one tiny bit. (Any job that can be done well by a machine should be, as far as I'm concerned.) But it does feel strange to think of a decent-sized virus being read off in that amount of time.
I started in the drug industry in 1989, so I span the pre- and post-genomics eras. The author George MacDonald Fraser pointed out that it was entirely possible for someone to be born in a covered wagon on the Oregon Trail, and then live long enough to watch a show about it on television. I feel a bit like that sometimes, so you can imagine how the folks feel who predate me. In the era that was closing when I came on, the genome was this huge, dark, pool, full of who-knew-how-many new receptors and enzymes. Estimates were all over the place for how many members of a given class might exist, and there was no way to prove any of them wrong.
If your drug for the Whatever receptor, subtype 2, didn't work they way you thought it should, there was always a good chance that it was simultaneously hitting the yet-unknown Whatever receptor, subtype 3. And those subtypes themselves had only been very recently defined from gene sequences. Before that, they were defined by what drugs seemed to hit them and what drugs didn't. As the genetic wave advance, some of these classification schemes proved to be pretty accurate, and others turned out to be full of holes.
As did all those estimates I spoke of. It's well known that the total number of human genes came out lower than almost anyone had predicted. That brought home to everyone how much variety comes in post-transcriptionally (and how much complexity and flexibility that builds in.) Now, we know that there are no more than a certain number of, say, G-protein coupled receptors out there in the genome. The frontier is closed.
But long live the frontier! To stick with the example of GPCRs, we've been finding out that what we thought were homogeneous populations of receptors really aren't. They can have different signaling pathways, depending on what proteins they associate with in the cell. They can bunch up, acting as pairs or multimers, with different properties than they would have alone. They can be mirror-images of their normal selves - instead of their signaling being normally turned off until an agonist molecule flips their switch, they can be normally on until an "inverse agonist" shuts them down. And even that on/off switch concept is too simplistic, as we come to terms with what it really means for a molecule to be a partial agonist.
Well, that's more detail than I was planning on, and I'll save discussions of receptor complexity for a future date. (Try to contain your excitement!) But the whole situation is an illustration of what science is about - we try as hard as we can to open the box, take out the gears and remove the mystery. But every time we do, we find another box inside, even more mysterious, with a lock on it like we've never seen before.
And we start in again. . .
Hold the Silicon
I had to fight off the temptation again the other day to put a silicon atom in one of my molecules. Admittedly, that's not an urge that strikes every chemist in the drug industry, but it strikes me every so often. A couple of times I've given in to it, when there's been a good fit for it (and it's produced active compounds, too.)
One one level, it's an understandable thing to try. Silicon's in the same group as carbon in the periodic table, and shares many of its properties (chief among them having four bonds in a tetrahedral shape.) Being one row down, it's larger and woolier than carbon, of course, but there are larger and woolier atoms that make it into drug molecules all the time. But there are reasons that you don't see silicon atoms all over the place in pharmaceuticals, because it's not a perfect replacement.
Yep, so much for all the old science fiction stories about "silicon-based life." Some experience in organic chemistry would have wiped out that idea from a lot of old stories (although Isaac Asimov used it once, as I recall, even though he knew better.) To pick one big difference, you can string carbon-carbon single bonds out until you just can't stand it any more. There's no thermodynamic problem; go crazy. But silicon-silicon single bonds actually get weaker the more of them in a row there are. That's a big problem for biomolecules, which range in size up past the rather large to the seriously gigantic.
Other bonds are weaker, too. A carbon-chlorine bond is medium reactive, but you have to beat up on it to displace the chlorine (as opposed to a carbon-fluorine bond, which you can whack on all day to no effect.) C-Cl bonds are solid enough for dichloromethane and chloroform to be common solvents. But silicon-chlorine bonds have some real zing to them, as anyone who knows the distinctive tangy reek of the chlorosilane reagents can tell you. Si-Cl bonds will start to go on you in moist air (such as is found in your sinuses.)
There's no good silyl equivalent to common functional groups from the carbon world, either. Alcohols become silanols, and are generally unstable. Carboxylic acids, aldehydes, and amides are wiped out by the fact that silicon doesn't form double bonds to oxygen very well at all. And it doesn't form multiple bonds to itself very readily, either, so you can forget alkenes, acetylenes, and (most importantly) all the aromatic rings like benzene, napthalene, and so on. Silicon analogs of such things are exotic beasts that get papers published about them just because they exist.
But all that doesn't mean that you can't work in a plain-vanilla silicon atom and get away with it (at least in the sense of making a stable compound.) Not too much is known, though, about what the body does with such things, which makes the toxicologists nervous. But it's been done. There are compounds that have gone into human trials with silicon atoms. The problem is, I'm not at all sure that any have made it out the other side.
So I'll hold off for now. I have a few other weird functional groups that I want to work into my structures, and it wouldn't be prudent to waste a good chance on a tricky atom like silicon. Now, sulfur, there's an atom no one objects to - well, except the smell - and I've got a couple of sulfur ideas that would make any enzyme in the world sit up and take notice. But I'll hold off on the details in case one of 'em turns into a wonder drug. . .
I mentioned the other day that companies have been trotting out all their existing antiviral drugs to see if any of them are effective against SARS. One of the obvious candidates has just suffered a blow: ribavirin doesn't seem to be effective in vitro.
That'll be news to the folks who have been bidding up the stock of ICN, the compound's maker. Ribavirin has been kicking around, in one form or another, for at least twenty years. Every time there's a new virus causing trouble, it gets tried, on the you-never-know principle. It's a nucleoside analog, in the same class as the first wave of drugs against HIV (remember AZT?) Those were already in the can as failed antivirals at the time, a situation very similar to today's - although I hope that the similarities to the HIV epidemic stop right there.
There's actually been some confusion about ICN and ribavirin. They have a subsidiary, Ribapharm, which has a deal with Schering-Plough to supply the compound as part of a combination treatment (with interferon) for Hepatitis C. The form of ribavirin that has been getting the SARS news, though, is formulated for inhalation (in a nebulizer) and that's still under the ICN banner. That hasn't stopped Ribapharm's stock from jumping around some days, though.
My belief is that we're getting close to the end of the time when a known antiviral treatment is going to turn up as a useful SARS drug. Give it a few more weeks. But after that, the odds start increasing that anyone press-releasing a SARS drug is seriously mistaken (or wants to make other people commit a serious mistake, like funding them.) Because by then, we're past the known agents, and getting an unknown one up to a point that it's seriously interesting will take years, not weeks. Caveat emptor
A Tangled Web
To give you an idea of how interlinked and multitracked the body's pharmacology is, consider the AT4 story. AT stands for angiotensin, known mainly for the blood-pressure regulating peptide angiotensin II. The enzyme that produces Ang II, angiotensin converting enzyme, is of course the target of the tremendously successful ACE inhibitors. And in recent years, ligands that bind directly to the receptor for Ang II (the "sartan" drugs) have been developed, and they're also doing well.
So the angiotensin story is all about blood pressure, right? Well, there are some other peptides - Ang III binds at the same receptors the first two do (and its purpose there is still rather unclear.) But Ang IV, which is a short fragment clipped out of Ang II by still other enzymes, is a different story. It binds to a totally different receptor, the AT4 subtype, and this is the subject of an interesting review in the latest Trends in Endocrinology and Metabolism.
AT4 receptors are found all over the place, including heart, kidney, and blood vessels - just the sort of thing to make you believe that it's another blood pressure regulator. And it could be, but the real story seems to be its distribution in the brain. It's scattered in a definite pattern, and seems to be involved in several areas of the brain that have been linked with memory. As it turns out, if you give rats the Ang IV peptide, they seem to perform better in the standard rat memory tests (like the Morris water maze, about which more at some future date.)
Giving the rodents the peptide isn't easy, though. Like most brain-active proteins, the only way to get it to the site of action is, basically, with an electric drill. The brain is fanatically picky about what it'll let in from the general riff-raff blood circulation, and makes a lot of its signaling peptides on the other side of the circulatory barrier.
The field got a surprise when the receptor was isolated and purified. It turned out to be something that was already know to a completely different field of biology: an insulin-regulated membrane aminopeptidase (IRAP.) I'm willing to be that most people expected it to be a good old-fashioned peptide receptor, of which the brain has scores; to find that the target what something that other people knew as an enzyme must have been quite a jolt. IRAP's also found in the places you'd expect to see insulin responses, like muscle and fat tissue, and it had been found to be a player in glucose uptake into such cells. Its fans must have been equally puzzled to find it playing a memory-enhancing role in the brain.
Actually, there's a lot of odd stuff going on with insulin and the brain that no one understands. There are plenty of insulin receptors up there, for one thing. That makes sense, in a way, because of the brain's dependence on glucose for fuel, but the receptors are localized in ways that make it appear that something else is going on, too. IRAP might be one key to figuring out what that is.
There are several things it could be doing up there. It's an aminopeptidase, of course, so it could be messing around with the formation or degradation of other signaling peptides. It could be involved in glucose uptake like it is in the periphery, and it could well have its own signaling pathways independent of these actions. So there are several major pathways, tangled up for us to unravel: blood pressure, insulin, and memory. And there's no reason that the complications have to stop there. That's a general principle, actually - there's no reason that the complications have to stop at all. . .
So Vaxgen (see this post) is being sued by investors. At least five law firms have filed class-action suits alleging that the company concealed the negative trial results from its HIV vaccine and thus deceived investors. The first suit alleges that the company's officials knew by early August that the vaccine wasn't working.
That would be quite a trick, since the data hadn't even been worked up from the clinical trial by then. Making a statement about the trial data before it's even been analyzed - now that would be worth suing someone about.
Of course, even if they'd somehow known instantly that the vaccine hadn't worked and press-released it that same afternoon, the lawsuits would have come down on them anyway. These things are an annoyance that you know is out there and you can't really do anything about, sort of like blackfly season in the north woods. I think that anyone who had hopes for dramatically positive results from Vaxgen's trial should sue whoever was teaching them about HIV and immunology.
The lawsuits also claim that the company is overdoing their subgroup analysis in order to keep their stock price up. Actually, I don't think I disagree - to my mind, Vaxgen is grasping at comparisons that might not even be real. But that means that they're simultaneously being sued for letting their stock price drop, and for not letting it drop enough.
The company, meanwhile, used a recent Keystone symposium to defend its contentions about the subgroup data. It'll take new data to really convince anyone, though - and who's going to pay to generate it?
To the coalition forces and to the Iraqi people. Not a single thing I wrote about back on March 16th (below) has come to pass, and it's hard for me to express how happy I'll be if that piece continues to slide into irrelevance.
Just to let everyone know, my intent is to publish something new here each weekday. (I didn't notice until this evening that today's post, below, hadn't made it up, but I should have the hang of things now.)
I haven't been covering the SARS story much yet. For more perspectives from the Blogosphere, see Living Code, Medpundit and the wealth of stuff over at Futurepundit. It's still to early to say just how bad a threat it'll be. (There's plenty of concern, though - for example, the American Associate for Cancer Research very suddenly canceled its meeting in Toronto last week. Too many attendees had expressed concern about returning to treat their immune-compromised patients after the meeting, and the whole thing was called off on about three days notice.) But I will mention something unfortunate from the perspective of a med-chemist: don't expect a small molecule against the virus (whatever it is) any time soon.
For a lot of reasons, "fast" isn't something we do really well. We can screen a lot of compounds in a very short time, true. Once in a very long while, something comes out of that process that only needs a year of work before it's ready to be tried as a drug candidate. Despite many innovations, there's still a lot of handwork involved in making new drug molecules. Then you start all the toxicity testing in animals, then you try it in the clinic - Phase I, Phase II, Phase III - and pretty soon six, or eight, or ten years have gone by.
Of course, it may turn out that some of the existing antiviral agents have some effectiveness. No doubt every company with an approved compound is running it past the candidate viruses just to be sure. But in this case, there just aren't that many good compounds out there, because antiviral therapy is a very difficult field for medicinal chemistry to handle.
Viruses have far fewer targets to attack than more complex organisms (if you want to call a virus an organism, that is - and that's more an argument about language than it is about virology.) And the ones that present themselves are often in difficult areas like protein-protein interactions. That's why everyone in the field looks for some sort of enzyme or receptor that the virus needs to infect cells or to replicate once they've done so. Those are feasible small-molecule targets, while most things having to do with coat proteins, subunit assembly, and the like just aren't. Not at the current stage of things, anyway. We're going to have to learn how to wrestle with those kinds of targets eventually, but they're mostly out of our reach at present.
That's where immunology comes in. You want something done about a protein, ask another protein to do it for you. Progress against SARS will almost certainly come from the vaccine end of things, as in most viral diseases. We can only hope that the virus turns out to be from a family that's amenable to our vaccine technology. . .and I'm not so sure about coronaviruses, the current front-runner. . .
Welcome to the new home of the blog formerly known as Lagniappe. Not long ago, the folks at Corante contacted me and asked if I'd be willing to move the site over and join their list of bloggers - and after looking things over, I decided it would be a good idea.
The name change is something I'd been thinking about even before the offer to move. I tend to get a steady flow of Google hits from people wondering just what the word "lagniappe" means. I don't think I enlightened them much. (The name came from the title of a column I used to write in my college newspaper. If someone offers me a regular newspaper Op-Ed column or something, maybe I'll revive it again!)
But the focus of the blog isn't changing at all. (There's a new post just below to prove it.) As longtime readers will know (hey, some have been with me as long as fourteen months,) I tend to write about drug discovery and medical news (and the IP issues and financial implications that they have) along with general thoughts about research. My opinions will continue to be freely ladled on top of each serving, and they're still worth just as much as you're paying for 'em. And never fear, there will continue to be occasional digressions into useful topics like which lab reagents smell the worst.
For new readers, I haven't actually covered that last one before. Have patience; we'll get to it. In the meantime, I'm glad to have you, and I hope you enjoy what you find. Comments of all sorts are always welcome to the e-mail address at left, unless, of course, you're the nephew of some recently deceased West African strongman or have a surefire proposal to lengthen parts of my body.
Liver Cells On Demand?
Geron recently announced that it has been able to differentiate human embryonic stem cells into what appear to be functional hepatocytes - the standard workhorse liver cell. As they're not shy about pointing out (see the press release,) this could have some implications for drug discovery.
We spend a lot of time worrying about what the liver is going to do to our compounds. Most pharmaceuticals aren't water-soluble enough to be given a free pass by the body's metabolic machinery, and the liver usually finds something on them to oxidize or break off. That's not always a bad thing, as long as it's a reasonably clean and controlled process. After all, it's better for your compound to exit the body in pieces than to never leave at all. One form of trouble starts when the changed metabolite that the liver forms turns out to be pharmaceutically active (or worse, toxic) all by itself.
Generally, you don't want to spur the liver to become even more aggressive about clearing out your compounds, but that's just what can happen. And it happens unpredictably: some compounds will cause the heptatocytes to produce the metabolizing enzymes in greater amounts. This enzyme induction can kill off a drug, because it can confuse the dosages of other drugs your patients might be taking. In extreme cases it can lead to liver damage.
The methods used to detect all these problems have problems of their own. The best ways are the ones that are closest to the real system, but human liver cells aren't easy to get. You usually can't culture them in the lab, and when you do they tend to lose their identity and many of their unique abilities. That leaves you with isolating them from the source, and live human liver tissue isn't exactly a catalog item.
Short of that, you can test your compounds against broken fractions of liver cells, which still have much (but certainly not all) of their metabolizing machinery intact. These can be frozen and used as needed, and are a key component of drug testing programs throughout the industry. As for enzyme induction, there are efforts to unravel the molecular mechanisms behind it, but that gets into the areas of nuclear receptor signaling and transcriptional control, two notorious hairballs. There's been a lot of progress in this area, but there are no assays yet that people are willing to risk their whole drug program on.
Geron's hope is to provide the drug industry with an industrial source of fully capable human liver cells, which could circumvent most of the bottlenecks I've described. But it's going to take some time to convince everyone that these cells do what they're supposed to do. As companies start to deal with Geron, look for them to get samples of the cells in-house and try out all their notorious failed compounds on them, the ones whose unexpected liver problems torpedoed their development. Every company has these things in their files, and they're just the kind of thing that everyone wants to avoid in the future. It'll be a tough test.
I should disclose that I'm a Geron shareholder, not that you'd necessarily be able to tell from the above. Unfortunately, I have a good amount of the stuff at long-ago (and fondly remembered) double-digit share prices, so the recent run-up in their stock, while enjoyable, doesn't even get me close to breaking even. And, speaking of that move in their shares, there was something rather suspicious going on in the last few days of March. Geron stock ran up on no news at all for a couple of days before this liver cell item was released, a move noticeable enough to be picked up as a news item all its own. Coincidence, no doubt. Or at least that's what some folks are probably going to be trying to tell the SEC. Good luck, guys!
Archives, RSS & Email
Click here for access to our archives.
Dictionary of Scientific Quotations
If the Universe is Teeming with Aliens
(Then where is everybody?)
The Night is Large
A Martin Gardner collection
Eurekas and Euphorias