The world seems to have come together in a perfect storm against Biodel (BIOD) whose NDA for Linjeta, a fast acting insulin, is under review by the FDA with a PDUFA date of October 30th. The stock closed today at 3.57, falling endlessly from its recent highs above 5.50 and not too far above its 52 week low of 3.22. As somebody on the Yahoo forums stated, “the charts show a toilet bowl formation occurring.” According to NASDAQ, as of the 15th there was almost 4.7 million shares short, almost 27% of the float. No doubt the number is even higher today. It also appears that sentiment on BIOD is at an all time low with the most recent comments, from an article by Brian Orelli at Motley Fool and two articles by G. Chambers at GekkoWire, suggesting that BIOD is unlikely to be accepted.
In dark times like these, it is important to go back and review the basics to see if any mistakes or oversights were made in our study of the data or our logic to prevent panic from taking over. It immediately helps to read these four excellent articles by Joseph Krueger and another article by Rockford Coscia. These two authors, in addition to G. Chambers, have written about this topic in more detail than I can possibly repeat here. The fact of the matter is, between the articles on Seeking Alpha and GekkoWire, you will find pretty much all the major knowledge on what was in the NDA. However, I will state my own major thoughts on this.
1. Linjeta is injected insulin. There is no doubt that it is safe. Insulin is broken down by the body from the hexameric form, which is used for storage, to the monomer before binding to the receptor. The purpose of Linjeta is to do this breakdown outside the body. The other ingredients in Linjeta, EDTA and citrate are also generally accepted as safe. EDTA is a chelator that will bind calcium and iron. It is possible that chronic high dosage of EDTA will cause hypocalcemia and anemia due to iron loss but the concentrations used in Linjeta, to my knowledge, will likely be too low to have any significant effect systemically. In any case, EDTA is thought to have a generally beneficial effect via its antioxidant effects. The “acid” used in Linjeta is citrate. This is a natural ingredient that is inside of most cells as part of the tricarboxylic acid cycle, the process cells use to release the energy in sugars. It is also a mild anticoagulant but it there is no question about its safety either since it is used as part of apheresis blood donation. If it is safe enough to give to the general public who are donating blood, it is safe enough to use in a patient as part of a treatment. There has been some concern about the recent rejection of Bydureon (exenatide), a GLP-1 analogue, in order to do QTc studies. Can the same thing happen to Linjeta? The answer is “possible but most likely not” due to the time courses involved. Bydureon, being a once weekly drug, has effects that last much longer than insulin, which prompted the QTc concerns because it may accumulate at higher concentrations. Linjeta is not a completely novel compound, so the FDA will not require such studies by default. It has a time course that is actually shorter than normal insulin, so if we agree that insulin is electrophysiologically “safe”, we must logically conclude that there is no concern regarding QTc with Linjeta. To summarize, there should be little doubt in anyone’s mind about the safety of the drug.
2. If we agree that Linjeta is insulin, we must also logically agree that it works as insulin. This is a conclusion that many people seem to get caught up on and this logical difficulty, plus the conundrum of the India trials, is what has sent this stock so low. By now, we all know that for Type 1 diabetics in India, there was some data anomaly with a batch that caused the entire group for Type 1 to fail the statistical test of noninferiority to Humalin but that it passed this test for Type 2 diabetics. Let us step back for a moment from the finer details about heat exposure, hemoglobin A1C, and the fact that both Humalin and Linjeta samples were adversely affected. Lets just think about whether it is possible for something that is “just insulin” to work for only Type 2 diabetics and not Type 1 diabetics. Type 1 diabetes is characterized by the failure of the pancreas to produce insulin. Type 2 diabetes involves a “resistance” of the tissues to the action of insulin after it binds to the receptor. One of the manifestations of this resistance is a decrease in the number of insulin receptors in Type 2 diabetics. If Linjeta is insulin, then if it fails in any population, it must fail in the Type 2 diabetics first and not the Type 1. Since the India study seems to show the impossible, that it works in Type 2 but not Type 1, only three possible logical explanations can be reached: that Linjeta works in a novel way that is not in common with insulin; that Linjeta does not work as well as insulin but the Type 2 data is wrong; that Linjeta does work as well as insulin but the Type 1 data is wrong. For the first explanation, we can say that it is likely untrue since Linjeta is insulin, as it would be seen in the body. The second explanation is also likely to be untrue since the Type 2 studies everywhere, assuming that they were adequately powered, showed that it was noninferior to Humalin. By a process of elimination, this leaves us with the third explanation. Here, we can finally bring in the additional evidence of the fact that the Humalin samples also seemed to be tainted, that physiological endpoints such as less weight gain and a trend of fewer hypoglycemic events still happened regardless of what the A1C said, and that it worked for Type 1 diabetics everywhere else. The NDA also included follow-up studies, and any continued similarity between Indiaand the rest of the world via their physiological end points will support our conclusion. Finally, although the FDA probably will not consider this data in their decision, Biodel has continued to study Linjeta’s effects in Type 1 diabetics and found that it had a faster onset of action than the fast acting insulin analogue lispro. They also note that “additional studies are planned to evaluate the impact of Linjeta on glucose control” so they have not redone what was missed in theIndia studies on another population of Type 1 diabetics. It is possible that the FDA will look at the phase 3 study and conclude that they don’t trust anything that Biodel reports to them because it is clearly impossible, but if they think about it logically, their most likely conclusion is that corrupted India data can be properly excluded. At this point, their decision to accept or reject Linjeta will be more testament to their trust in Biodel’s integrity and management.
3. The citrate in Linjeta’s original formulation is necessary for the breakdown into monomers but the low pH caused pain on injection. Linjeta’s proposed final formulation involves a neutral pH to eliminate pain. Another question is whether the FDA will buy the argument that this new formulation is equivalent to the old one. If you look at the Linjeta vs lispro press release, they also studied the differences between the formulations with different pH values and found that they were bioequivalent. There has been some argument that since low pH was used to break down the insulin hexamer, that restoring the pH would cause it to reassemble, negating the purpose of Linjeta. While this is chemically possible, it is also unlikely. Insulin is a complex peptide that required special enzymes to assemble. Once the peptide is broken, it would be very unlikely, without help from specialized enzymes, for the pieces to reorient themselves just right to reform the hexamer. Think of Linjeta as a steak (insulin) dropped in acid (citrate). After the steak breaks down, just because you titrate the acid back to neutral, the steak will not spontaneously regenerate itself. A fundamental law of nature, that entropy always increases with every process, prevents this from happening spontaneously. Finally, one should not be overly disturbed by the headaches reported in the press release. Headaches are a common result of insulin use due to fluctuating glucose levels and in any case, we do not have any idea of whether this occurs more frequently than other insulin types.
4. Another thing we should consider is how has the FDA historically looked at NDAs for insulin? I looked on the FDA website and unfortunately, you cannot download reports for drugs approved in the 90s or before so humalin, velosulin, NPH, and lente cannot be analyzed. If you consider the NDA for lantus, there was a concern about increased progression of retinopathy in type 2 diabetics, but this was not cause for rejection. For Levemir, the phase III study did not even show noninferiority for white type 2 diabetics, which makes much more logical sense compared to noninferority just for indian type 1 diabetics, and the medical reviewer even recommended that it be rejected pending more studies on type 2 diabetics. However, the decision was still made to accept the NDA. There did not appear to be any issues with Apidra. It seems that, at least historically, the FDA has been willing to accept some errors in the NDA, including a failure to show noninferority in a phase III trial. The FDA has become more stringent on safety recently but have they also become more stringent on proof of efficacy? I guess we will see.
5. At this point, we have thought hard about the science of Linjeta and what the FDA thinks about applications for insulin. Despite reaching some logical conclusions, it is still quite possible that we are wrong because we don’t know all the facts or our judgement is still somehow unsound. There are some people who may know more of these facts than us or may be better decision makers. What are they doing with biodel? Biodel is 30% owned by insiders, as opposed to 1% and 2% respectively for recent failures such as ARNA and JAZZ. Note that AMLN had a healthy 19% insider ownership but their CRL was truly unexpected. On October 1st, Biodel’s CEO reduced his monthly salary to increase his pay in restricted stock units. This is not the activity of a person who expects his stock to fall. Note also on October 5th, a total of 8 insiders reported receiving restricted stock. Whether this is a normal part of their compensations process or something of their own doing is impossible to determine but it is certainly not a bad thing. Another event is the recent acquisition of a 5% stake on October 12th by BAM Capital. Probably the strangest feature is that as of the end of last year, BIOD was the 2nd largest stock holding of Yale University (note that Yale does not own much stock directly so it is a tiny portion of their endowment). You can also take some (although not much) comfort in that a total of 50% of 206 voters on the GekkoWire expected either a full or type 2 approval on BIOD. Although there is some positive activity on this side, it is very difficult to see who has been shorting BIOD. Perhaps they also have more knowledge of the facts? Unfortunately, there is no way to tell. Thus, the information presented in this section has strong affirmation bias and should be taken with a grain of salt.
Of course, as was seen in many recent decisions, it is not reasonable to expect anything close to perfect accuracy in predicting the FDA. The scary thing about these decisions is that they really can go either way depending on the mood and whims of the committee members. However, despite heavy shorting of BIOD and some sense of panic among longs, no material facts have changed. I believe Linjeta still has a strong logical case for acceptance.
Disclosure: Long BIOD and BIOD calls