I wrote about the Celldex data in a story posted Wednesday night. What I like about the CDX-011 data were the positive signals of efficacy in breast cancer patients who over-express the GPNMB protein and in triple negative disease. The latter group of breast cancer patients is particularly important because of the complete lack of treatment options currently.

CDX-011 is an antibody targeted at cancer cells that express GPNMB so tumor response that is directly correlated with increased levels of GPNMB expression makes a lot of sense. The analysis of patient subgroups around GPNMB expression levels (5%, 25% and 50%) and triple negative disease was prospectively defined before the study was started. That's important.

Legitimate concerns and/or stuff we don't know yet that is potentially risky: 1) Small numbers of patients, especially when looking at some of the subgroups. You have to be worried that the positive signals seen in the study may be diminished or disappear in a larger study with more patients.

2) Celldex hasn't disclosed response rates in patients with greater than 50% GPNMB expression levels. The company told me that results were fairly consistent and suggest that higher GPNMB expression levels resulted in greater response rates. That's fine but Celldex still won't tell us the response rate in the third bucket of patients with greater than 50% GPNMB expression levels.

I want to see those data because if what Celldex says is true, it will further verify the mechanism of CDX-011 and increase the chance that this is real drug. Celldex may be reticent at this time to give us these data because the patient numbers are too small, but it's still important information and somewhat bothersome that the data are being kept from us.

Let me be clear before you start yelling at me: I'm more optimistic about Celldex and CDX-011 today than I was before the data were released. The company deserves kudos for the way it handled Wednesday's announcement. But that doesn't mean CDX-011 is a guaranteed success. Far from it, we've still only seen a sliver of data that leaves important questions and risks on the table.

Stay tuned for more CDX-011 data in December, likely at the San Antonio Breast Cancer Symposium.

Sticking with Celldex, @ColfaxCapital asks, "Adam, do you think CDX-011 is a better drug than $GALE's breast cancer drug?"

Yes, from the data we've seen so date, CDX-011 has more potential as a breast cancer drug than Galena Biopharma's NeuVax vaccine. One caveat, the two drugs are very different and are being developed in different breast cancer patient populations.

Jay R. writes, "Adam, I am a big follower of your writing and generally I am fan of your reporting on biotech companies. Your article today about Celldex surprises me. While there are indeed 'silver linings' in these data, the company was clearly data mining. As you know, this is a process by which companies sort through their clinical data to find the best possible story to tell investors and the FDA. While it is logical that these patients would do best, if the company felt so strongly then the trial should have been prospectively designed to only look at these patients. In addition, there are only 35 patients in the analysis of triple negative, high expressers. When Keryx Pharmaceuticals ( KERX) tried to pump their perifosine colorectal results, you cautioned investors about the small size of their trial. Shouldn't you be consistent here Adam?"

Jay's concerns are certainly legitimate, particularly about the small numbers of patients in the CDX-011 story. I raised the same issue above. Celldex, however, did design this study prospectively to enroll patients who over-express GPNMB. They also pre-defined, or stratified, patients into three big buckets based on the level of GPNMB expression (5%, 25% and 50%). For this reason, I don't have concerns about the cutoff points of GPNMB expression used in the analysis and I don't consider it to be data mining.

Keryx, on the hand, absolutely data-mined positive results from its phase II study in order to come up with a colon-cancer patient population to study in the phase III trial. The result was predictable -- a failed phase III study.

Speaking of Galena, Avi B. writes, "So you are promoting Celldex with a 19% effectiveness versus 14% control group study on all patients in the trial and yet you continue to write negatively against Galena who's results were much stronger. CDX-011 has shown significant improvement in a subset of patients, so does NeuVax, yet you discount the data presented by Galena and tout Celldex. Your agenda is now evident and your reporting lacks professional consistency. "

If you liked this article you might like

Biotech Movers: OncoMed Tanks After Barrage of Bad News, Akorn Rises on Deal Talk

Medivation Founder Takes Helm at Axovant, Shares Spike

Second-Quarter Picks and Pans

How I'd Play Biotech Buyout Rumors

How I'd Play Biotech Buyout Rumors