About this Author
DBL%20Hendrix%20small.png College chemistry, 1983

Derek Lowe The 2002 Model

Dbl%20new%20portrait%20B%26W.png After 10 years of blogging. . .

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: Twitter: Dereklowe

Chemistry and Drug Data: Drugbank
Chempedia Lab
Synthetic Pages
Organic Chemistry Portal
Not Voodoo

Chemistry and Pharma Blogs:
Org Prep Daily
The Haystack
A New Merck, Reviewed
Liberal Arts Chemistry
Electron Pusher
All Things Metathesis
C&E News Blogs
Chemiotics II
Chemical Space
Noel O'Blog
In Vivo Blog
Terra Sigilatta
BBSRC/Douglas Kell
Realizations in Biostatistics
ChemSpider Blog
Organic Chem - Education & Industry
Pharma Strategy Blog
No Name No Slogan
Practical Fragments
The Curious Wavefunction
Natural Product Man
Fragment Literature
Chemistry World Blog
Synthetic Nature
Chemistry Blog
Synthesizing Ideas
Eye on FDA
Chemical Forums
Symyx Blog
Sceptical Chymist
Lamentations on Chemistry
Computational Organic Chemistry
Mining Drugs
Henry Rzepa

Science Blogs and News:
Bad Science
The Loom
Uncertain Principles
Fierce Biotech
Blogs for Industry
Omics! Omics!
Young Female Scientist
Notional Slurry
Nobel Intent
SciTech Daily
Science Blog
Gene Expression (I)
Gene Expression (II)
Adventures in Ethics and Science
Transterrestrial Musings
Slashdot Science
Cosmic Variance
Biology News Net

Medical Blogs
DB's Medical Rants
Science-Based Medicine
Respectful Insolence
Diabetes Mine

Economics and Business
Marginal Revolution
The Volokh Conspiracy
Knowledge Problem

Politics / Current Events
Virginia Postrel
Belmont Club
Mickey Kaus

Belles Lettres
Uncouth Reflections
Arts and Letters Daily
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline

In the Pipeline

« Compounds and Proteins | Main | GeneVec's Pancreatic Cancer Therapy Crashes »

March 29, 2010

Antisoma's Phase III Disaster

Email This Entry

Posted by Derek

We get reminded again and again that interesting Phase II results are only that: interesting, and no guarantee of anything. Antisoma (and their partner Novartis) are the latest company to illustrate that painful reality - their drug AS404 (vadimezan) looked in Phase II as if it might be a useful addition to oncology treatments, but has completely missed its endpoints in the bigger, more realistic world of Phase III. The trial was halted after an interim analysis showed basically no hope of it showing benefit if things continued.

There are many reasons for why these things happen. Phase II trials are typically smaller, and their patient populations are more carefully selected. And they're quite susceptible to wishful thinking. They're designed to keep things going, to show some reason to proceed, and they often do. If your drug candidate makes it through Phase II, that may say more about how you designed the trial than it says about the compound.

That's not to say that getting past Phase II is meaningless. Compared to having no efficacy data at all, it's a big step. But Phase III, when a compound goes out to a larger and more diverse patient population, is a much bigger one. And plenty of candidates aren't up to it.

Comments (28) + TrackBacks (0) | Category: Cancer | Clinical Trials


1. darwin on March 29, 2010 9:16 AM writes...

Anyone have any idea what the ratio of internal Phase III to in-licensed Phase III failures are?

Permalink to Comment

2. MTK on March 29, 2010 9:54 AM writes...

A couple of personal opinions here:

a) all clinical trials, Phase II or III, are designed to support the desired labeling. Study design, particularly in selecting the patient population, is done in every trial. Even Phase III trials generally do not match how the drug will be used in real life.

b) The success rate of clinical trials in any phase is all over the map depending on your source. I assume that also depends on your definition of success. The failure rate in Phase III should be lower than in either Phase I or II, you would think. The cost of Phase III trials is too high. You want to minimize failure in Phase III. Regardless, of what the absolute numbers are, it seems that most of the estimates have Phase III success rates that are equal or slightly better than Phase II.

Darwin, see Nat. Rev. Drug Disc. 2005, 5, 451. Upshot: In-licensed compounds have higher overall success rates, but fail more often in late stages. Since late stage failures are more costly than early stage, it's probably close to a wash economically.

Permalink to Comment

3. Anonymous on March 29, 2010 9:54 AM writes...

Is this related to AS1404 that Novartis bought with $890 mm in 2007?

Permalink to Comment

4. Sili on March 29, 2010 9:56 AM writes...

Anyway to design Phase II to be less useless?

Or is deceiving the investors a too much of a feature?

Permalink to Comment

5. David P on March 29, 2010 10:01 AM writes...

@4: the same thought occurred to me: surely it is better to get the bad news at phase II than at phase III? Can they not design trials for phase II more like what their phase III trial will be like? Or is that just crazy talk?

Permalink to Comment

6. Ed on March 29, 2010 10:08 AM writes...

Isn't this just business as usual for UK biotech?

Permalink to Comment

7. CM on March 29, 2010 10:10 AM writes...

Clearly it's better to succeed in Phase II, even if it fails spectacularly at Phase III (Dimebon anyone?). As it will take years to subsequently fail at Phase III at that point it's somebody else's problem.

Permalink to Comment

8. lynx on March 29, 2010 10:15 AM writes...

I'm no chemist, but that looks like one ugly molecule to me. Is it more than an intercalator (I did a bit of research and as far as I can see it's vascular disrupting mechanism is still unclear.) I'm amazed there is any specificity at all.

Permalink to Comment

9. sgcox on March 29, 2010 10:45 AM writes...

#3: yes.
Novartis must be really desperate to fill its pipeline. compound was first first published as anticancer agent in 1990 ! And now, almost 20 years later makes headlines. Novartis, discover of Gleevec... Any progress since ?

Permalink to Comment

10. skatesailor on March 29, 2010 10:49 AM writes...

Efforts to discover and develop drugs resemble adultery in three respects. "Adultery," said H. L. Mencken, "is time consuming, expensive, and likely to be humiliating."

Permalink to Comment

11. T on March 29, 2010 11:19 AM writes...

#9 I thought the NV pipeline was one of the stronger in the industry? They certainly don't seem to be cutting back on R&D like some others.

Permalink to Comment

12. Sally on March 29, 2010 11:30 AM writes...

Advanced lung cancer is particularly difficult to treat given the life expectancy is only about a year to begin with.

Earlier this year both Pfizer's figitumumab and Novelos' NOV-002 both failed in front-line phase III trials in non-small cell lung cancer. Pfizer also abandoned Sutent in the indication last year.

#9 Novartis have launched Zometa and Afinitor since Gleevec and have others in the cancer pipeline too. These results are very disappointing indeed :(

Permalink to Comment

13. sgcox on March 29, 2010 11:46 AM writes...

#12 yes, it is certainly better then most of the others, but both examples are hardly breakthrough innovations: Zolendronate is bisphosphonate (discovered in 19th cntuary) and itself was known since earlier 90, Everolimus is a small modification of rapamycin - known since 1975.
Granted, Novartis made Nilotinib - but again is a imatinib follow-up.

Permalink to Comment

14. alig on March 29, 2010 11:59 AM writes...

Bear Stearns published a report in 2008 that showed internal Big Pharma Phase 3 success rates were close to 90% whereas in-licensed (post-phase 2) programs had only a 50% phase 3 success rate.

Permalink to Comment

15. RTW on March 29, 2010 12:01 PM writes...

I knew this as DMXAA. It was probably at the time we continued to still have a strong collaboration with Professors Baguley and Denny prior to the Pfizer take over of WL/PD. This is just one of many compounds investigated by these two oncology researchers at the University of Auckland/The New Zeland Cancer Society. They have a very good research group there that does drug discovery, preclinical and I also beleive clinical oncology research. We worked closely with them on several projects. I was pretty unhappy when Pfizer decided not to renew the collboration agreement after the takeover. They where a great group of people to work with!

Permalink to Comment

16. Hap on March 29, 2010 1:20 PM writes...

#14: Is there something in Bear Stearns's analysis that is consistent with Morgan Stanley's analysis? If the dependence of research failure rates on research source is so strong, then how did MS come up with the idea that research should be outsourced, rapidly? If the gap were narrower, MTK's analysis would be a likely cause for MS's analysis, but with that large a gap (and the back-loading of costs in clinical trials), the two pieces of research don't appear consistent.

Permalink to Comment

17. MTK on March 29, 2010 1:54 PM writes...


That Bear Stearns study number you cited are not an indictment on in-licensing, but rather more indicative on how difficult it is to have a coherent clinical plan.

If you look at the numbers, the overall failure rate of in-house compounds is worse than in-licensed compounds. In-licensed ones, especially those brought in late, have a high failure rate in Phase III. A clinical development plan is really tough.

Phase III plans need to answer specific questions in specific numbers in order to get approval with the desired labeling. Phase II studies need to answer those same questions, or surrogates, in sufficient numbers to justify the risk of going ahead with Phase III. When you have a change in ownership during or after Phase II, there's bound to be some level of disconnect resulting in a disjointed clinical plan and therefore a greater risk of failure.

There's a reason why few small companies develop drugs themselves. Besides the expense, clinical trials and clinical programs are very difficult and require a huge amount of expertise and skill to design, execute, interpret, and report.

Permalink to Comment

18. Anonymous on March 29, 2010 2:47 PM writes...

"When you have a change in ownership during or after Phase II, there's bound to be some level of disconnect"

The disconnect being that a biotech Phase II is designed to relieve senior pharma execs of their money whereas all Phase III programmes have to be aimed at a meaningful effect in the marketplace.

I doubt very much whether the failure is ever due to a disjointed clinical plan. More often I think it's due to the inability to hide the warts once you get into Phase III.

But let's get real here - anytime anyone analyses a process and tells you that something is twice as good as what you are doing (in this case in-licensing vs internal R&D) I guarantee they are talking bulls**t. Tell me ten percent and I might believe you. But the headlong dash by the herd down this path will necessarily alter the market - simple supply and demand. Plenty of opportunity then to sell more rubbish to the gullible until the money really runs out.

Permalink to Comment

19. CMCguy on March 29, 2010 3:23 PM writes...

#4 & #5 while I am sure every Clinician would love to get all the answers in Phase II it is a difficult and rare task to even get that far. Phase IIs are building on Safety info from Phase I and while indeed design and expectation is this will guide what to do in Phase III there is often many consideration that complicate matters. Confounding factors makes medchem look trivial by comparison so would you argue that supporting R&D deceives investors when discovery can not just create the right drug from the beginning?

All Phase IIIs are disappointments, not only for companies involved but think about the patients. The info may be somewhere but I did not see what the endpoints that were missed are. Because this is an add on treatment my assumption is that had to run as superior to current regime and may well have shown promise in Phase II but in Phase III the statistics did not hold up (meaning it could be working just not to the extent needed to gain approval by the predefined criteria). Oncology sometimes benefits from better translations but can suffer for greater disease diversity. NSCLC is notoriously difficult, but has great potential market, so is best not to generalize but often in such failures there are a few minor variations that could have occurred to have made pathway to success more favorable. Problem is one does not realize them at the time and usually a few years later in retrospect that positive and negative decisions become evident.

Permalink to Comment

20. MTK on March 29, 2010 3:28 PM writes...

Anon #18,

That would be a cynical way of looking at it.

Just about every clinical plan is designed to give the drug the best chance to reach the market. That's regardless of whether it's Big Pharma, Biotech, or whatever. That doesn't mean the studies are rigged or that they are not legitimate, it means things like identifying the proper patient population, comparators, clinical endpoints, etc.

A perfectly fine Phase II program which gives great results for a compound may result in Phase III failure if the clinical endpoints or the patient population changes, for example. As an example, let's say you have a hypertension drug which rings all the bells in Phase II for patients with extremely high blood pressure. The original intent was to apply for that labeling and the Phase III study was going to use that particular patient population. Big Pharma X buys it from you and decides that in order to make it a real blockbuster, they're going to include patients with even mild hypertension. Boom, the trial fails. That's not due to any duplicity on the small company nor due to any "warts" in the compound. That's a disjointed clinical plan where the Phase II studies did not support the Phase III studies. Big Pharma X took a flyer.

Or how about a osteoporosis drug where increased bone density is the primary clinical endpoint. It is a reasonable measurable endpoint where significance can be reached in a reasonable trial size. Big Pharma X comes along, however, and would like to have labeling supporting lower fracture incidence. That's fine except that the Phase III for such a study would take more patient-years than originally planned. X decides to go ahead with a smaller (or shorter) than optimal study. It ends up falling short of significance. Once again a disjointed clinical plan.

These things happen all the time, regardless of whether in-house or not, but it's certainly bound to happen more often when different organizations are involved.

The lesson, IMO, is if you're going to in-license then do so early. If you in-license late, then make sure that the clinical plan and results match your goals for the project.

Permalink to Comment

21. gippgig on March 30, 2010 1:44 AM writes...

There is an interesting article ("Odds Are, It's Wrong") about problems with statistics in the latest (March 27) issue of Science News magazine. The article frequently refers to drug studies and may be relevant here.

Permalink to Comment

22. medchem23 on March 30, 2010 2:42 AM writes...

One of the reasons internal phase III studies may be more successful is that they are often the culmination of a series of cycles between the labs and the clinic for a particular target. This cyclic assessment and redesign gets broken down by adopting an in-licensing strategy and all the requisite knowledge required to achieve clinical success becomes very diluted or lost.

Permalink to Comment

23. milkshake on March 30, 2010 3:00 AM writes...

...but having left their roots, and themselves given to the stream’s tyrannous rage, alas, are driven
through mills, and rocks, and woods and at last, almost consumed in going, in the sea are lost.

Permalink to Comment

24. Anonymous on March 30, 2010 7:52 AM writes...

"That would be a cynical way of looking at it."

But then I've sat in the opposite position as a scientist begging our senior management not to buy in a phase II candidate (thereby killing our internal phase I and preclinical compounds) simply because the data looks nice. I've seen some real horrors bought in with nasty structures where the internal experts were simply over-ruled. And where are those compounds now - dead or dying, along with the perfectly decent internal compounds that they took out when they were bought in.

Now $800 million here, $500 million there, before you know it you're talking about real money.

Permalink to Comment

25. Edward Taussig on March 31, 2010 8:42 AM writes...

One of the clearest warning signs that a Phase III trail might fail is when the Phase II trial didn't even reach statistical significance, and only via post-hoc analysis are meaningful results claimed for sub-groups. Look at the results today from Arqule for ARQ 197 for a perfect example. Odds are they'll move it to Phase III, it may work out in the long run, but the risks are *much* higher when a drug doesn't meet statistical significance for the initial criteria.

Permalink to Comment

26. alig on March 31, 2010 11:51 AM writes...

Apparently the CEO of ArQule doesn't read your blog:
"Phase III only puts the finishing touches on the drug," asserted CEO Paulo Pucci

Permalink to Comment

27. Hap on March 31, 2010 12:04 PM writes...

Apparently he doesn't read this blog, or pay attention to newspapers, or read any business magazines in his company's area of expertise, or...

I hope that's strategic incompetence on his part and not something else. I also hope that his stockholders/investors aren't dumb enough to fall for that line.

Permalink to Comment

28. Edward Taussig on March 31, 2010 5:50 PM writes...

Another warning sign is when the goal of a phase II trail is to identify effectiveness versus a biological "marker",
a marker which has not yet in itself been proven to have a casual relationship to the disease that is the real target of the drug.

Even if a subsequent phase III trail confirms the effect against that 'marker',
there's no guarantee it will have any disease modifying effect.

Permalink to Comment


Remember Me?


Email this entry to:

Your email address:

Message (optional):

The Last Post
The GSK Layoffs Continue, By Proxy
The Move is Nigh
Another Alzheimer's IPO
Cutbacks at C&E News
Sanofi Pays to Get Back Into Oncology
An Irresponsible Statement About Curing Cancer
Oliver Sacks on Turning Back to Chemistry