Telik, and a transaction or two...

Telik announced on the 25th of September that their 3rd line ovarian cancer trial, ASSIST-1, had hit the prespecified number of events. The primary endpoint of this trial is overall survival / reduction in risk of death, similar to ASSIST-2 which had hit its prespecified number of events back in the spring.

So both A-1 and -2 are ready to analyze. ASSIST-3, the combination trial in platinum refractory / resistant ovarian cancer, had fully enrolled in the late spring. The primary endpoint of this trial is response rate, with a secondary endpoint of progression free survival (PFS). Since it has been roughly 5 months since the closure of that trial, it is relatively safe to assume that patients have had 6 cycles of drug on either arm, if necessary (drug is administered every 3 weeks). Therefore, there has been enough time for responders to respond, and nonresponders have likely progressed by now, especially since median PFS in this population is around 4-5 months.

In short, all the trials are ready for analysis. If we are to assume that it would take 4-8 weeks until the results are released, investors are looking at a pleasant or unpleasant surprise soon (it has been 3 weeks since they noted A-1 had hit the number of events).

Since part of this exercise is to see how well we do at predicting drug success, I'm going to put down a prediction and a few words. We can come back to it in a few months and have a good chuckle. We'll rank the trials from most likely to hit primary endpoint, to least likely:

Most Likely to Hit Primary Endpoint: ASSIST-3

This trial is looking at response rate for patients taking carboplatin and telcyta versus those taking Doxil. Doxil's response rate is pretty well confined to ~15%, whereas carboplatin and telcyta has shown a response rate over 40%. Provided that tumour responses are independently verified, they are probably the most direct indication of a drug effect and most reliable in terms of reproducibility. Whereas other endpoints such as PFS and survival can be greatly influenced by reporting intervals and the heterogeneity of patients and their stage of disease / post-trial care, response rate isn't as influenced by these factors. That said, I do think that A-3 is likely to reproduce the 40% or so response rate seen in the phase II trial, and will show a statistically significant improvement over Doxil.

Also, as the response rate approaches 50%, I do think that there is a very decent shot at showing an improvement of PFS in this trial. My only reservation for the PFS measurement of this trial is that the data may not be mature if they locked the database right after the last enrolled patient was given a chance to show a response (after 5-6 cycles of drug). If there are many patients like this, then their PFS would be unknown / censored. If there are too many censored patients on both arms, then it may reduce the ability to see any stat sig difference in PFS.

Really, this trial's primary endpoint makes its success pretty much guaranteed. It's the regulatory relevance of this trial that is up for debate.


Next Most Likely to Hit Primary Endpoint: ASSIST-1

This trial is looking at overall survival in 3rd line ovarian cancer. The median overall survival data from the previous phase II trials have looked good when compared to Doxil / Hycamtin, and the confidence in those data is increased because two such trials were run. Small biotechs often run one phase II trial and then shuffle off to phase III, so running two phase IIs allows investors to get a better feel of whether or not the results are reproducible. In both trials, the response rates were reproducible, suggesting that the drug effect is there. Whether that will reproducibly result in an improvement in survival is the question. My feeling is that there is a good chance of hitting the primary endpoint, although this trial is not the gimme tha A-3 appears to be.

Least Likely to Hit Primary Endpoint: ASSIST-2

In the two phase II trials that preceded A-2, Telik saw response rates of zero and 11%. The trial with no responses ended up showing a median survival greater than 10 months for the 41 patients reported. No data were released for the trial with the 11% response rate. Obviously if the 10 month value could be repeated, this trial would be a success. But since one trial had responses and the other had none, the reproducibility comes into question. NSCLC is notorious for showing phase II trials with good median survival that drops like a rock in randomized trials. This trial also has an international flavour, which can always be a red flag / wildcard. All in all, I think this is the one most likely not to succeed.


Regulatory Relevance?

Success in A-1 and/or -2 will be sufficient for FDA approval, in my opinion. Both trials are looking at overall survival / reduction in risk of death and both trials have SPA agreements with the FDA. If either hits the primary endpoint, I would think that approval is assured (barring unexpected and unforeseeable circumstances).

There is the issue of whether A-1 had an interim analysis, and how this may or may not have influenced the statistical hurdle at the final analysis. The company has mentioned that this trial had the chance to support accelerated approval, and that if any interim data suggest a course change, investors will be notified. Investors weren't notified of anything, which doesn't mean that there were or were not interim analyses. To be honest, it is unclear to me what types of interim analyses Wick talks about in this regard. Trials do have safety boards that convene and intermittently look at the data to determine if one arm is being unduly risked in the trial. These types of analyses are not the formal "interim" analyses that a company and the FDA negotiate ahead of time, and which necessitate a statistical penalty. However, the ambiguity of the comments and the mention of accelerated approval leaves open the possibility that there may have been a comparison of objective response rate and / or PFS at some interim point. If it was response rate, it would not be surprising if the analysis did not stop the trial. If it was a PFS analysis, it would be somewhat disappointing. This is an outstanding issue that I've not been able to completely clarify, and remains a risk.

The A-3 trial is the one that has the least regulatory relevance. I do think there is reasonable debate as to whether or not a trial with response rate in second line ovarian cancer will be sufficient, by itself, for FDA approval. This question mark becomes larger since the scenario that would necessitate A-3 being the registrational trial for approval would necessarily mean that A-1 failed. It would be difficult to argue for approval of Telcyta based on a surrogate endpoint when a trial testing survival fell short.


The All-Important Predictions


1) Assist-3 will hit the primary endpoint. No doubt.

2) My best guess is that both A-1 and -3 will end up hitting their primary endpoints whereas A-2 will not. This scenario would be sufficient for approval of Telcyta. The market reaction may be muted, because NSCLC is the larger market that one wants to enter. However, the triplet data of telcyta + carbo + taxol in NSCLC is looking good, and this is probably Telik's best shot at entering the NSCLC market in an appreciable manner. Competition in 2nd/3rd line NSCLC is fierce, and unless Telcyta has blow-out results in A-2 (unlikely), they would be unable to compete significantly against the sharks from DNA / Lilly / Sanofi even if they showed an increase in survival over Iressa / placebo.

So if we place some percentages, I'm looking for success in A-1 and A-3 as the most likely outcome (45%). This is followed by success in A-3 alone (40%), then A-2 and A-3 (10%), and least likely scenario, A-1, 2 and -3 (5%). You'll note that I do not think any scenario suggesting failure of A-3 is possible, which removes a whole host of scenarios including the deadly "each of A-1, 2 and 3 fail".


On other matters, we're going to make a couple of additions:

1) Purchase 1200 shares of IDIX at Friday's closing price of $9.00, total cost of $10800. The PDUFA date is the end of October, and I think approval is likely. Going to buy the stock in anticipation of (cross our fingers) earnings. That leaves us with cash of $38672.62.

2) We're also going short 2800 shares of INGN at Friday's (the 13th!) closing price of $4.4701, a total "cost" of $12516.28. This company is simply not being straightforward with the market, so we're going to take advantage. Watch for their Advexin registration strategy to fall flat on its face. I'm looking to cover this at around $3.50 per share, barring unforeseen developments. The general strategy with short positions will be to hold long term even if it goes against us in the short term. Since we're mostly concerned with fundamentals, and since INGN's fundamentals suck, patience should provide good returns. This technically returns our cash position to $51188.90, but I will continue to consider the cash at $38672.62 because I do not want to have to mark things to market, calculate margin requirements, etc... (see #1 below).

I did have one suggestion to put the leftover cash into HQH, a dividend paying "fund" that is filled with public and private biotechs that are the focus of a blog like this. It's a good idea, but I'm not going to do it for a couple of reasons:

1) I'm lazy. I don't want to follow the quarterly distribution. (Yes, this is a very weak reason)

2) More importantly, I want to keep the cash available because there may be a time in the future where I may want to take advantage of an unfavorable market in biotechs to pick up some shares that I may consider to be bargains. If I keep the cash on hand in HQH, then an overall decline in biotechs will cause a fall in HQH and will lessen the buying power especially at the time when I may want that extra buying power. Clearly I could try and time an exit out of HQH if I feel the tide is turning, but that becomes an even more expansive market-timing exercise that I don't want to overly burden myself with.

It may also occur to some that I could put the remaining cash in a money market fund and maximize the return from the cash, albeit modestly, while it waits to be used for future purchases. This is true, but see #1 above... I'm lazy. I don't want to calculate average balances etc... At least we know that the performance of this portfolio would be relatively "modest" in comparison to what one would truly get in the real world since their broker could easily track the cash balance and provide investors that little extra yield. Of course, I don't include commissions which would reduce the yield, so maybe all these things balance out in the long term.


Also, in case people forget: Please assume that I have a short or long position, as you deem appropriate, in any company mentioned here. Assume the most sinister possibility because that is fine with me. I'm not one of those people who thinks that such general commentary on the web truly moves stock prices. And if you're going to email me or post comments with conspiracy theories on why I have this blog and the positions that are disclosed, please make them creative.