Celldex's triple-negative breast cancer progress

Triple-negative breast cancer (TNBC) is a nasty form of the disease that does not respond to receptor-targeted therapeutics (Herceptin or Tamoxifen), as the receptors of interest (estrogen (ER), progesterone (PR), or HER-2) are not found in TNBC. The only marginally-effective treatments against TNBC are general chemotherapies, but overall response and survival rates are much lower in TNBC versus other breast cancers.

Celldex (CLDX) released P2B results from a clinical trial using their drug (CDX-011) in TNBC. Results suggest that the Celldex drug works against triple-negative cancers that overexpress a protein known as GPNMB. Celldex reports that 36% of patients with TNBC and high GPNMB expression responded to CDX-011, while zero patients in the equivalent control groups responded to standard chemotherapies. (N=11 and 3 respectively, so let's not overreact.)

(Also: NON-TNBC patients with high GPNMB expression showed a response to CDX-011. (32% response vs. 13%. N=25 and N=8, respectively.

Adding these two groups together yields a response in 15 of 36 patients (42%!) with high GPNMB expression treated with CDX-011, with 1 responder of 11 in the control group. (i.e. not treated with CDX-011.)

This data is encouraging for Phase 2B, and warrants a Phase 3 trial AND a large pharma partner, something I would expect to see Celldex close on by the end of 2012. (TNBC sometimes responds to EGFR treatment, so I'd expect Celldex's partner to be a company with an EGFR product interested in prescribing a combination of products. (Hello, AMGN, and Roche!)

(Speaking of combination, CDX-011 uses Seattle Genetics' antibody linked technology, whereby the antibody with affinity to GPNMB delivers a chemo payload to the cancerous cell. I need to take another look SGEN soon.)


CLDX Financial overview, as of 5/28/12:
Market cap: $264M (even after popping ~10% on the tnbc trial news
Cash on hand: $92M, burning ~$40M/yr.
Enterprise value: $172M
ex-CDX-011 enterprise value: $35M-$50M, as CLDX receives $9M/year in license income from successful outlicensing.
CDX-011 value (roughly): $130M ($172M less $42M in ex-CDX-011 enterprise value ($42M= midpoint of $35M-$50M valuation.)


What's the value of CDX-011?

CDX-011 Market math:
CLDX asserts that 35% of all breast cancer patients could benefit from CLDX-011, which suggests:

Annual US breast cancer cases: ~180,000
% of breast cancer that is TNBC: 15-25% via various sources. Let's say 20%.
US TNBC market: 36,000
GPNMB overexpressers as a %age of TNBC patients: 12% of TNBC, or 4,320

US GPNMB market:27,000
non-TNBC patients overexpressing GPNMB: 15% of all breast cancers
US market for GPNMB overexpressors (w/o TNBC): ~27,000 
Likely US GPNMB+ market: 31,320 annually, and therefore,
Total annual world market for CDX-011: between 80,000 and 120,000 patients worldwide.

Assuming $20k revenue/patient ($50k/patient revenue, but only 40% penetration at peak), CDX-011 would then peak at ~$2B/yr in revenue, and therefore peak product value of ~$12B (equivalent to Celgene's 6X sales valuation.)

Risk-adjustment of CDX-011 value: let's assume the following:
-that FDA approval is 2 years away,
-peak revenue is 4 years post-approval
-25% discount rate
-CDX-011 has a 40% chance of successful P3 trials....

Based on peak market value of $12B for CDX-011, you could then project that CDX-011 has a present value of $1.25B, or, with a current valuation for CLDX of $172M, that the market thinks that there is only a 14% chance that my scenario above becomes true. Either way, the likelihood that Celldex is undervalued is high.

(btw: is use the value of CLDX and CDX-011 interchangeably. While CLDX does have other products in the pipeline, 99% of CLDX valuation will depend on CDX-011.)

(CLDX says that 35% of breast cancer patients could benefit from targeting GPNMB with CDX-011. It's probably a rosy-case press release figure, but even if you chop this number in half, the resulting NPV for CDX-011 3.6X today's valuation.)


Factors for Celldex:
1) Few, if any, other GPNMB programs in existence, so it is a seller's market (when it comes to partnering) for Celldex.
2) There's plenty of reason to expect CLDX to partner CDX-011 soon, and it's an ideal time to partner - P3 trials costs are significant and therefore better shared with a partner, plus partnering now allows the Big Pharma partner to influence trial design.
3) CDX-011 clinical responses exceed the 30% hurdle rate and in P2B trials have a wide advantage versus the control arm. (I'm always skeptical if the response rate and advantage versus control group outcomes is <=20% and <=10%, respectively.)
4) Good sized market, not currently served.


Factors against Celldex:
1) small-cap biotechs NEVER successfully get drugs to the market by themselves.
2) ultimate FDA approval may be conditional for GPNMB+ only, and also could depend on the development of a gene-specific test.
3) lack of partner to date could represent Big Pharma skepticism over either GPNMB's target biology or Celldex's capabilities.
4) Limited ability to raise equity financing.
5) small sample size in P2B results.
6) time to FDA approval of ~2yrs limits the ability to buy call options for CLDX.

Ultimately, the value of CLDX is driven by 1) CDX's long term outlook, and 2) the size and timing of a partnership with a big pharma to commercialize CDX-011. Outlined above is one long-term scenario representing a win for CLDX, and for the other point (partnership value), I can easily see CLDX doubling the company's valuation this year after consumating a big pharma partnership centered on CDX-011. (Guess: $50M cash @ closing, $50M in equity purchased by the pharma partner, with $1B in milestones possible; also: $50M in easily achievable milestones to land in the first 12 months. (example: $25M payment at initiation of P3 trials.)

Here's hoping that CDX-011 lives up to the P2B results, and that finally we'll have a weapon against TNBC.


Disclosure: at the time of writing, I DO hold a tiny position in Celldex.

ex-Pfizer R&D head vs. bank analyst on drug discovery strategy: who ya got?

Forbes magazine unintentionally hosted a good drug discovery strategy debate. It started with a prominent pharma industry bank analyst Jack Scannell critiquing therapeutic R&D productivity. His points: 1) targeted drug development has been less productive than other approaches, and 2) high-throughput R&D technologies really haven't been productive either.

John LaMattina, formerly Pfizer's head of R&D fired back, also in Forbes ("Analysts get it wrong again"), which attributes lower R&D productivity to.........pharma mergers and more demanding regulators and payors. (Never mind that increasing R&D productivity has been the rationale for much of the industry consolidation.)

Both make good points, though. HTS and genomic technologies have definitely under-delivered. But, while the industry in the early days of HTS and genomics truly WAS guilty of treating drug discovery as a numbers game, researchers have become much smarter more efficient in their use of these technologies. (Whereas some R&D centers initially built labs to maximize compounds screened per day ("100,000 per day capacity!"), most are using HTS (and other technologies) to more inexpensively examine smaller focused libraries.)

Note: neither side cites budgets (neither pharma nor NIH) as an inhibitor of R&D productivity.

Scannell says that the numbers don't lie - 33 of the 50 first in class drugs studied started from a phenotypic-centric philosophy, but LaMattina counters that this is explained by the lag inherent with tech adoption, and that a wave of targeted compounds is on the horizon.

This is tough analysis to choose a side on - I think the phenotypic approach has been the benefit of low-hanging fruit (i.e. development to date has benefitted from easy molecules, but there aren't nearly as many easy ones left), while the targeted approach just has an inherent intellectual appeal. ("If we know what causes disease "X," why not just target it?")

(That being said, one of the more significant tech flops of the last decade or so has been "Rational Drug Design.")

I'd also nominate one other reason for low R&D productivity not mentioned by Scannell or LaMattina: organization structure. Innovation becomes the exception and not the rule as organizations grow bigger, while risk tolerance seems to decline. That bigger organizations stifle drug development is reinforced by the notion that many of the successful therapeutic programs were once considered UNsuccessful programs, as LaMattina's story of the invention of Viagra indicates. Another reinforcing story is that of Gleevec's development from Daniel Vasella's book: only the singular efforts, passion,  and strength of Dr. Brian Druker kept a Novartis committee from killing off the lead that became known as Gleevec.

Let's hope that pharma R&D rises soon, whether because pharma mergers have slowed, or because productivity is catching up with the technology.

Uh-oh. (New cancer biology understanding to negate targeted RX?)

A NEJM article coming in 3 days is reported to prove that cancerous tumors are not monolithic in their genomic profile - to the extreme that multiple samples of the same patient's tumor express different genetic mutations, with only limited commonalities among samples from the same tumor. (Early news coverage here (WSJ), here (Bloomberg), and here (FierceBiotech).)

Implications: this is going to turn some worlds on their heads Here's my quick guesses at implications:

• Cancer just became even harder to solve. Think chess is complex? How about 3-D chess? That's pretty much the leap in complexity that cancer researchers just experienced. 

• Possible boon for DNA sequencing: demand for multiple sequences per patient may make DNA sequencing a bigger market faster. The NEJM study suggests that sequencing a tumor longitudinally (i.e. at a regular schedule, during treatment) will guide multiple treatment decisions, which may differ based on new expression patterns or mutations. Does this also mean that each tumor will be sampled many times in different sites at diagnosis? If so, we may move from 1 sequence/patient to dozens of sequences per patient. (Good gosh that's ALOT of data. 1 terabyte per sequence is hard enough to handle. 20 TB/patient? Wow.)

• Anyone working on hazily predictive PGX might be wise to give up. The idea of a single analyte for a single disease, or a number of analytes against a number of diseases probably only targets one portion of a given disease. In other words, if your predictive test isn't 100% predictive, you are only explaining a fraction of the disease, and therefore of negligible utility.

• This might be the death of Affymetrix.  I can't see much value from a highly variable 1-dimensional expression profile, unless they invent a way to generate multiple expression profiles from a single sample at roughly the sample price point.

• This news might actually boost sales of targeted therapies in the near term. Why test for HER-2 status? Even if a patient is tested HER-2 negative, an oncologist wouldn't be out of line to still treat with Herceptin, knowing that the HER-2 negative status may only apply to a portion of the tumor.

• Along the same lines, the NEJM finding may give a boost to combination therapies. For example, Sutent, Nexavar (VEGFR and PDGFR) and Raf kinases), Torisel (mTor), Votrient (VEGFR-1, VEGFR-2, VEGFR-3, PDGFR-a/β, and c-kit), and Inlyta ( VEGFR-1, VEGFR-2, VEGFR-3,platelet derived growth factor receptor (PDGFR), and cKIT (CD117).) are each sold for RCC (kidney cancer) but each seem to be only marginally effective. They have different inhibitory profiles though - could combining them produce a better outcome? Roche would seem to be in the best position to gain if combinations are successful.

• Would a need for combination therapies be the straw that breaks the camels's back on the American health care system or even the USA? If treatment with a single targeted therapeutic may be $30,000 per month, would a combination therapy be $100k/month? I can't imagine Medicare and private insurers are ready to pay these sums. If they are, Medicare is already a multi-trillion dollar liability. Wanna go for quadrillion dollar liability?

• This may be a boon to systems biology researchers, like Lee Hood and his team at the ISB. Hood has for a long time seen cancer not as a product of a single mutation, but rather a cascade of biological signals which in total result in cancer. 

• Expect more attention to early detection and treatment. Cancers are FAR less complex in their earlier stages of development.

• Just a guess on my part: more surgical biopsies, less needle biopsies if more tumor material is needed?

• The result: we need more different cancer meds to mix and match patient profiles. Could the FDA loosen up a little on approval requirements, or would this make it even more difficult, as any new targeted drug for a given disease would need to succeed or fail in combination with other targeted therapies? (i.e. more false negatives a false positives from combinatorial effects.)

• Good news/bad news. Bad news: every single clinical-stage targeted therapy just became less valuable. That drug targeting gene "X" is less valuable, now that gene "X" explains less of the disease. Good news: targeted drugs that narrowly failed late stage trials might be resurrected. Maybe the drug didn't fail because it wasn't effective against the target, but rather because the target explains less of the disease than previously believed.

Ultimately, this news dampens the optimism for personalized medicine, but because of cancer's proven ability to mutate, most of us knew that cancer would not be beaten by single silver bullets.

Xcovery blog revisited (state of targeted Rx)

About five years ago I started a blog dedicated to targeted therapeutics, especially kinases inhibitors. The blog was an outgrowth of Xcovery, the kinase discovery spin-out from the Scripps Research Institute that I started and served as EVP of Business Development. 

I was already tracking developments in biopharm so the blog was an outlet for some of basic analysis and a fun way to share my opinion and connect with others in the industry. 

One of the regular bits of analysis was tracking the performance of FDA approved targeted drugs. Just for fun, here's a five year update, with some analysis:

 

Of note:

• The 17 approved molecularly targeted drugs accounted for $27B in global sales in 2011. Think about that for a second, then consider that most of these drugs have been on the market for only 5-6 years, and their approved indications are still growing. Consider too that most have not been applied as combination therapies.

• Even the senior citizen of the group (Herceptin, approved in 1998), has seen prolonged growth, averaging 36% per year over the last five years.

• With 8 blockbusters and several more close and still growing (Tasigna, Sprycel, etc), almost all of the targeted drugs are either blockbusters, or well on their way. So much for the concern that targeting drugs might limit the market potential.

• The top 4 (Avastin, Herceptin, Gleevec, and Lucentis) have made a mockery of their projected sales ceilings and are still growing strongly.

• On the other hand, the only assets that appear to be underperforming expectations are Amgen’s Vectibix, GSK’s Tykerb, and Pfizer’s Torisel (specific sales data isn’t available for 2011, as Torisel is listed under “other oncology,” totaling ~$130M across several drugs.)

• Vectibix is still playing catch up to Erbitux, and Tykerb hasn’t gained much traction against the Roche juggernaut.

• I wonder what Amgen’s new CEO will do about Vectibix. It seems that there’s 2 choices: go big (invest in expanding trials for more indications and in comparison with Erbitux) or go home (sell the product to another biopharm.)

• 4 of the top 6 are Roche drugs, which means that they were discovered by Genentech. Hats off again to the DNA team in South San Francisco for their amazing science and productivity. I wonder if we will ever see any other drug discovery effort be so inventive and productive for a prolonged period.

• Also: I don’t think anyone is doubting the wisdom of Roche buying the piece of DNA that Roche didn’t own. I haven’t run the numbers, but I’d be shocked if the DNA acquisition wasn’t a resounding financial win for Roche.

• Unfortunately, OSI’s acquisition of Macugen was a tremendous dud.

• I am encouraged by the progress since my last analysis in 2006 – an average of two new approvals each year, with most new products addressing new targets or diseases, in contrast to the incremental “me too-ism” in other pharma areas like ED or cholesterol drugs.

A few sweeping generalizations:

• FDA approval and sales success seem to be connected to corporate resources. Small to mid-cap biotechs have been chasing targeted therapies for ~15 years without much output. (I’m talking about companies such as Exelixis, Vertex (pre-HepC), Ariad, etc., though I don’t mean to pick on specific companies.) With three exceptions (Onyx’s Nexavar, OSI’s Tarceva, and the former ImClone’s Erbitux), the targeted therapies have largely been developed in-house by “old” companies with multi-billion dollar market caps and the resources to match. (You could make the case that Amgen’s Vectibix came from a small targeted effort at Abgenix, but I suspect that it was Amgen’s resources that got Vectibix through FDA approval. Similarly, Sutent started at Sugen, but Pharmacia and Pfizer seemed to have provided the big push.)

• A gross generalization: the small to mid-caps tend to lack broad biological or disease-specific expertise, instead investing in target-specific expertise, or platform-specific expertise, thinking that broad expertise (ancillary to their target or disease of interest) is expensive overhead. I wonder if the results to date argue for the big pharma discovery model, or just reinforces the need for a broad portfolio to be successful in drug discovery and development.

• With rare exception (as in Pfizer’s Xalkori and Novartis’ Gleevec), the path to FDA approval has been arduous for these drugs. There are a number of targeted drug developers who hold out hope that their P2 or P3 results will be so clear and strong that their clinical trials will be stopped early and approved quickly. That’s definitely the exception, unfortunately, and even in the positive trials for targeted drugs, the data has tended to be good, not great. I suspect that is a function of the requirements of clinical trial design and comparison to first-line chemotherapies. As a result the “new” drugs are posting smallish survival benefits when compared to the “old” therapies, with no accounting for how certain patient segments have had dramatic benefits. (Thus starting the vicious circular argument that targeted therapies ought to have stratified patient populations in clinical trials, but stratifying patients shrinks the market potential for such drugs, bring the business viability of the targeted therapy into question.) It seems that the FDA could take the Xalkori experience and develop a novel process for rapid approval based on patient stratification without derailing or obviating more broad approval for the drug.

The $27B in revenue in this segment (likely to grow past $50B in 2014) has hopefully served to further de-risk pharma R&D in molecularly targeted therapeutics. Coupled with advancements in medicinal chemistry, we will hopefully see more and better targeted therapies in the future.