NEW YORK (TheStreet) -- Another column in which I talked to myself, this time about MannKind (MNKD) and the inhaled, rapid-acting insulin Afrezza. An FDA-sponsored advisory panel convenes Tuesday to review Afrezza and issue a recommendation on whether or not the product should be approved for the treatment of Type 1 and Type 2 diabetes. The following made-up conversation previews tomorrow's Afrezza panel.
Adam, you previously gave 60% odds for a negative panel vote on Afrezza, but that was before the FDA weighed in with its clinical review on Friday. Where do you stand now on Afrezza's chances after reading the FDA's briefing materials.
The same 60% chance of a rejection by the FDA advisory panel. I'm more negative now on Afrezza in Type 1 diabetes -- I believe MannKind is in serious trouble there. For Type 2, the debate and vote will be closer. MannKind has a chance but a lot has to go right for them.
Going into the release of the FDA briefing material, many MannKind critics, including yourself, questioned whether the regulators would conclude that one or both of the Afrezza phase III studies had actually failed -- contradicting the claims made by the company. That didn't happen, so isn't MannKind's outlook for tomorrow better?
FDA didn't come out and say the Afrezza studies failed. That's true, but regulators weren't exactly effusive in their praise of the data, either. FDA described Afrezza's blood glucose-lowering efficacy as "modest" several times in its review while also raising questions about interpretations of the efficacy results.
For example, In the Type 1 diabetes study, the FDA raises concerns about background basal insulin levels that were significantly higher in the Afrezza arm compared to the control arm. Normalizing those basal insulin levels could have turned a narrowly positive study (Afrezza just barely squeezed by the statistical non-inferiority margin) into a negative one.
In the Type 2 study, FDA agreed with MannKind that Afrezza was superior to placebo for the primary endpoint of HbA1c reduction, but the agency also went out of its way to include "comparative effectiveness" data showing how Afrezza was inferior to almost all currently available Type 2 diabetes therapies.
Then, FDA raised concerns about Afrezza's lung safety, including a small, but higher incidence of lung cancer seen in Afrezza patients. Modest efficacy plus safety concerns is not a favorable balance for a drug seeking approval in a chronic condition like diabetes, especially one where patients have an abundance of effective medicines already available.
You admit MannKind has a decent chance to emerge from tomorrow's panel meeting with at least one positive vote recommending Afrezza's approval. What does the company need to do to make that happen?
As best it can, MannKind needs to keep discussion focused on the Afrezza data from the two studies, emphasizing the company did what FDA asked for in order to get its product approved. It's hard to imagine any of the panelists tomorrow being wowed by Afrezza, but if MannKind can demonstrate the data are good enough for approval, it may get a positive vote.
MannKind may have to concede away approval in Type 1 diabetes and also be very acquiescent about the need for additional, post-approval studies to better quantify Afrezza's long-term lung safety.
The day goes bad for MannKind if panelists or the FDA make a stink about the clinical utility of Afrezza relative to other diabetes therapies, or if the company comes across as argumentative or inflexible.
There is no FDA webcast of Tuesday's panel but you're live-blogging, correct?
Ah, thanks for the self promotion. Yes, I will be live-blogging the Afrezza panel. You'll be able to find a link to the live blog on TheStreet's home page, starting Tuesday at 7:45 am EDT. If you want to watch the panel yourself, FDALive.com provides a live web feed but the cost is $200.
Will MannKind trading be halted Tuesday?
I don't know for sure but I'd assume, yes, MannKind will not trade during the Afrezza panel meeting.
What time will the panel members vote on Afrezza?
You can download a copy of the agenda for Tuesday's panel. It's an all-day affair. The panel starts at 8 am EDT with introductory remarks by FDA. MannKind will present the Afrezza data to the panel first, followed by the FDA's presentation. If you can't log in to my live blog all day (and why not?!?!), the most important part of the meeting will take place probably between 3 pm and 5 pm EDT. That's when the panel members are expected to debate the Afrezza data and vote on whether or not to recommend approval.
MannKind supporters dislike you and believe you're biased against the company. Why should they bother to log into your live blog when all you're going to do is bash Afrezza and the company. Basher!
Not true! Well, I'm not a MannKind fan, that is true, but I've tried to make factual and fundamental arguments for why Afrezza is not the miracle insulin therapy its supporters believe it is. Disagreement is a good thing. With that said, I promise to give MannKind a fair shot during my live blog. You can ask fans of Amarin and Arena Pharmaceuticals how I treated the companies during their respective FDA advisory panels. They may still dislike me, but they'd also tell you, begrudgingly, that I was fair.
Let's get back to discussing the Afrezza studies. You mentioned above that MannKind might be in trouble with the Type 1 diabetes indication. Can you explain why?
Sure. The phase III study in Type 1s compares Afrezza plus basal insulin against insulin aspart plus basal insulin. (A third arm of the study uses the older "Medtone" inhaler but I won't discuss that here.)
At the end of the study, the mean decrease in HbA1c from baseline for Afrezza was -0.21 compared to -0.40 for insulin aspart -- a difference of 0.19. From a statistical standpoint, Afrezza was found to be non-inferior to insulin aspart because the upper bound of the confidence interval (0.36) was within the agreed-upon margin of 0.40.
That sounds like a lot of gobbledygook, but what you're saying is the Afrezza study in Type 1 diabetics was successful.
Yes, the FDA agreed that Type 1 Afrezza study met the primary endpoint of non-inferiority, but...
Uh oh, there is a but?
There are a few buts, none are positive. At all time points in the study, Afrezza's ability to lower HbA1c is numerically worse than insulin aspart. FDA also notes that in a previous phase III study of Type 1 diabetics, Afrezza (using the older Medtone inhaler) was found to be statistically inferior to injected insulin with an almost identical between-group differences in HbA1c seen in the new phase III study.
What's the significance?
The FDA is saying when you look at the two phase III studies in type 1 diabetics, Afrezza came out numerically worse compared to injected insulin both times. The latter study was deemed positive only because of the non-inferiority statistical analysis.
And then there's the basal insulin analysis.
Yes, another potential problem in the Type 1 indication. FDA notes that Afrezza users required more basal insulin across the entire study compared to insulin aspart patients. Here's what the FDA reviewer concluded:
The Afrezza T1 Gen2 group's use of more basal insulin compared with the insulin aspart group is further evidence that Afrezza TI Gen2 is less effective than insulin aspart, at least as a substitute for prandial subcutaneously injected insulin...
After noting the difference in dose titration between the two arms of the study, the FDA reviewer adds:
Taking these observations together, it is not clear how to interpret the results from this non-inferiority study. It is concerning that if the insulin aspart group had been titrated more effectively, differences in efficacy between Afrezza TI and insulin aspart might have been greater and the non-inferiority margin may not have been excluded...
So, the FDA is saying, it's possible the Type 1 study failed.
Sort of. The FDA describes a plausible scenario under which the Type 1 study of Afrezza could have failed, but even under the most ideal circumstance, the efficacy of the product is modest. (There's that bad word, again.)
And what about the Type 2 study?
MannKind has a stronger case with the Type 2 study, which compared Afrezza against placebo, with all patients on background oral anti-diabetic medications. Mannkind and FDA agree the study achieved its primary endpoint, with mean reduction in HbA1c from baseline of -0.82 for Afrezza compared to -0.42 for placebo. Statistically significant.
Excellent for Mannkind!
Good for Mannkind, but...
C'mon, another but!
I know, but FDA did something very unusual and potential troubling for Mannkind in its review. The reviewer included a chart which compared the placebo-adjusted 0.40 reduction in HbA1c for Afrezza against the placebo-adjusted HbA1c reductions demonstrated by currently approved Type 2 medicines like AstraZeneca's (AZN) Onglyza and Johnson & Johnson's (JNJ) Invokana.
This "comparative effectiveness" doesn't put Afrezza in a good light. Here's the FDA reviewer's comment:
While cross-study comparison is difficult and generally not recommended, on its face, the magnitude of the effect size achieved with Afreszza TI in comparison to these other antidiabetes agents is surprising modest, especially in light of the fact that Afrezza TI can be titrated.
Up top, I said the key to Mannkind's winning Tuesday would be keeping discussion to its own data and trying to downplay any comparison of Afrezza with currently approved diabetes medications. You see what I mean now.
Should people be freaked out about the FDA raising concerns about a lung cancer risk in Afrezza patients?
FDA found a numerical imbalance in reported lung cancers going against Afrezza, but the incidence rate is small and animal studies showed no cancer signal. Insulin is a growth factor and a similar lung cancer concern was raised about Pfizer's (PFE) discontinued inhaled insulin Exubera, so I can understand why this will be a significant topic on Tuesday. However, it doesn't seem like a knock-out punch against Afrezza.
The bigger concern for Mannkind are the questions raised about overall lung safety and Afrezza, particularly bronchospasms in patients with COPD or asthma, and the unknowns around decreased lung function tied to long-term Afrezza use.
In an acutely fatal disease, none of these safety issues would be a deal breaker, but diabetes is a chronic condition with generally safe medicines already available. The FDA makes a point to say that the phase III studies under review aren't long enough to answer all the safety questions about Afrezza. Will the panel want to see more safety data prior to an approval, or will post-approval follow-up studies be sufficient?
The final decision to approve or not comes down to a risk-benefit equation: Do the benefits of Afrezza outweigh its risks.
Exactly, ultimately, that's the question the panel members Tuesday will debate and vote on. The FDA, in its review, doesn't draw a firm opinion, but seems genuinely interested in what the panel members have to say.
But anyone reading the FDA's review objectively should walk away with concerns because Afrezza is simply not all that effective relative to injected meal-time insulin. I'll end this too-long preview with the conclusion reached by the FDA's statistical reviewer.
The primary analysis from the T1DM (Study 171) met the criterion that TI (prandial insulin) delivered via a Gen2 inhaler [Afrezza], was non-inferior to insulin aspart in lowering HbA1c after 24 weeks of treatment in subjects whose disease were sub optimally controlled with theire current basal insulin regimens. However, the comparative efficacy shown here was not compelling since the upper band (0.37%) of the 95% CI of the treatment difference (TI Gen2 minus insulin aspart) in change from baseline in HbA1c at week 24 was almost right at the boundary of the pre-specified margin (0.4%), and the mean reduction in the TI-Gen2-treated patients was actually statistical significantly worse (by an estimate of 0.22%) when compared with that in the insulin aspart-treated patients...
Data from the T2DM trial (Study 175) have demonstrated that TI, delivered via a Gen2 inhaler, was statistically superior to placebo in lowering HbA1c after 24 weeks of treatment in subjects whose disease were sub optimally controlled on optimal/maximally tolerated doses of metformin only or 2 or more OAD agents. However, the treatment difference (TI-Gen2 minus placebo) in change from baseline in HbA1c at week 24 was modest (-0.4%)...
The final conclusions for approval of the drug/device should also take the comparability of Ti and insulin aspart doses as well as safety factors such as hypoglycemia and lung function into consideration.
Get a good night's sleep tonight. See you tomorrow for the Afrezza FDA panel and live blog.
Follow Adam Feuerstein on Twitter.